Independent oversight for the AI shaping work, money and rights.
AI WatchDog monitors where AI is deployed, records incidents with graded evidence, tracks regulation across jurisdictions and measures what AI does to workers, creators and markets. Organisations get the governance tools to act on it.
No pay-to-rank, everPrivacy by designPublished methodology
What is AI WatchDog?
AI WatchDog is an independent intelligence and accountability platform for AI's impact on people, markets and society. It combines first-person evidence from affected workers and creators, independent monitoring of companies and governments, and practical governance tools. Its public-interest core is free. Organisations pay for compliance, assurance and benchmarking. It starts with a Pakistan, Gulf and Global South lens and is pro-accountability, not anti-AI.
Platform modules
106features across 10 pillars plus certification
Launch scope
21critical features in the 6–9 month MVP
Evidence states
6from self-reported to independently verified
Cost to individuals
$0workers, creators and the public never pay
AI is moving faster than the institutions meant to watch it
The headline numbers are mostly US-centric and based on what employers and news outlets choose to say. Almost no equivalent evidence exists for Pakistan, the Gulf or the freelance economy.
121,401
US job cuts where employers cited AI, January to October 2026, about 20% of all announced cuts.
Challenger, Gray & Christmas, as reported Nov 2026
~18,900
AI incidents and hazards catalogued, drawn mostly from English-language news.
OECD AI Incidents & Hazards Monitor, Nov 2026
$1.5B
Class settlement over pirated books used for AI training, as courts begin to price creator rights.
Bartz v. Anthropic (US)
2 Dec 2027
New date for EU AI Act stand-alone high-risk obligations after the Digital Omnibus deferral.
EU Digital Omnibus on AI, provisional agreement May 2026
Figures as publicly reported up to November 2026. Re-verify before publication; update from the research team's source log.
One evidence base, sixteen ways to act on it
Every module draws on the same verified data: deployments, incidents, workforce outcomes, spending, creator rights and regulation. Filter by what you need to do, then open a card for details.
The public-interest core stays free
These programmes are why the platform exists. They give affected people a place to be counted and supply the first-person evidence no one else collects.
Count Me In: the worker registry
People whose jobs, income or tasks changed because of AI can register privately, follow up at 3, 6 and 12 months and be counted in anonymised totals. Other contributing causes are recorded too, so AI is never assumed to be the only cause.
WF-01, WF-02, WF-03, WF-04, WF-09
Creator Rights Registry
Writers, artists, musicians and performers record their works and training-use preferences, look up whether their work appears in known datasets, and follow AI copyright cases in plain language.
CR-01, CR-02, CR-06, CR-11
AI Spend & Adoption Observatory
A recurring survey of what organisations spend on AI, where they deploy it and what it changes. It feeds sector benchmarks and the annual State of AI Impact report.
EC-01, DI-04, DI-08
Community, Academy & newsletter
Moderated forums by sector and city, resilience circles for people in transition, AI literacy modules in English and Urdu, and the Weekly Watch Bulletin.
CE-01, CE-02, WF-04, PA-04
Built for everyone in the AI impact chain
People affected by AI, the institutions that decide how it is used and the organisations that build it all need the same evidence base, just in different ways.
Allegations are never presented as established facts. Each record carries one of six evidence states, a source trail and a confidence label. Named organisations get a right of reply before or alongside publication.
Self-reportedSubmitted by a participant and shown as testimony, with consent
Document-supportedBacked by submitted records; evidence type and review scope shown
CorroboratedMultiple independent sources; usable in aggregates with caveats
Independently verifiedDocumented review; scope, date and reviewer role published
Disputed or under reviewMaterial disagreement; shown as a neutral status only
Corrected or withdrawnChanged or removed, with correction history kept where lawful
See what your board and your regulator will see
KPI cards, compliance scorecards, risk heatmaps and incident registers in one workspace. Open the full interactive preview.
An independent intelligence and accountability platform that monitors where AI is deployed, records incidents and harms with graded evidence, tracks AI regulation and measures AI's effects on workers, creators, consumers and the economy.
Is AI WatchDog against AI?
No. It is pro-accountability, not anti-AI. It counts harms and gains, recognises good employers and offers solutions such as reskilling and rights guidance alongside criticism.
How are reports verified?
Every record carries one of six evidence states, from self-reported to independently verified. Disputed records are shown neutrally, and named organisations get a right of reply.
Can a company pay to change its ranking?
No. Commercial relationships never affect rankings, incident records, Watch List entries or index scores. Accurate findings are never removed for payment.
Which regulations do you track?
The EU AI Act and Digital Omnibus, US federal and state laws, Pakistan's National AI Policy 2025, UK and Gulf rules, and voluntary frameworks including NIST AI RMF, ISO/IEC 42001 and UNESCO's Recommendation.
I lost work because of AI. How do I register?
Join the Count Me In registry for free. You choose what to share, only anonymised totals are published, and you can withdraw at any time.
Bring evidence to your next AI decision
See the Governance Workbench, regulatory intelligence and benchmarking in a 30-minute walkthrough tailored to your sector.
Sixteen modules in four groups, all running on one shared data model with one verification process. Each module lists the backlog features it is built on and the launch phase it ships in.
The full roadmap covers 106 features in eleven groups. The modules below draw on several pillars at once. Each feature code, such as AT-04, starts with the code of the pillar it belongs to.
WFWorkforce Impact Observatory
Worker registry, layoff and AI-washing checks, reskilling, freelancer income
ECAI Economics & Spend Intelligence
Spend and adoption survey, pricing transparency, ROI, public procurement
CRCreator Rights & AI Copyright
Rights registry, "Was My Work Used?", case tracker, licensing pilots
Forums, AI Literacy Academy, fellowships, open data commons
PAPublic Awareness & Advocacy
Human Stories of AI, Weekly Watch Bulletin, media desk, campaigns
DIData Insights, Indexes & Benchmarking
Dashboards, indexes, rankings, State of AI Impact report
CTCertification & Standards
Assurance, seals and certifications, introduced only after safeguards mature
Observe
Know where AI is deployed, how it changes and what goes wrong.
AI Monitoring
MVP
AI Monitoring keeps a continuous public watch on AI systems used in hiring, credit, health, public services and consumer products. Each system is recorded as a plain-language Transparency Card.
AI System & Use-Case Watchlist, searchable by sector, country, provider and risk tier
Transparency Cards covering purpose, limits, data, human review, appeal routes and evaluation coverage
Public-sector deployment map with procurement records from the Procurement Monitor
Change log that flags edits to providers' terms, privacy and usage policies
For
Public, regulators, journalists, compliance teams
Built on
AT-04, AT-05, EC-04
Phase
Launch / MVP, months 3–9
Transparency Card example layoutIllustrative, not a real system
System and use
Automated CV screening for call-centre hiring
Deployment context
Employment, Pakistan; EU AI Act high-risk category
Provider disclosure
Partial Purpose and data summary published; evaluations not published
Human oversight
Recruiter reviews rejections; to be verified
Appeal route
Email to HR; no response time stated
Known limitations
Lower accuracy on Urdu-language CVs (provider statement)
Incidents on record
1, under review
Right of reply
Sent; response pending
AI Model Tracking
MVP
AI Model Tracking records which models and versions sit behind a product and how their policies, evaluations and capabilities change over time.
Model, product and policy change log with version timelines and diff views
Aggregated model and data cards with rights annotations for training sources
Urdu and regional-language benchmark leaderboard for quality and bias
Evaluation coverage gaps flagged against the use cases a model is sold for
For
AI buyers, researchers, journalists, creators
Built on
AT-05, EB-02, CR-02
Phase
MVP, then Growth (months 9–18)
AI Safety Monitoring
MVP
AI Safety Monitoring tracks the harms that spread fastest: deepfakes and AI fraud, AI agents acting beyond their authority, risks to children and frontier-model safety commitments.
Deepfake, synthetic media and AI fraud alert network with verification tools for the public
Autonomous AI agent incident watch for unauthorised purchases, data access and actions
Child and youth online safety monitor for chatbots and companion apps
Frontier AI safety watch and AI cybersecurity and supply-chain briefings
For
Consumers, banks, telecoms, parents, regulators
Built on
RM-03, RM-05, RM-06, RM-07, RM-08
Phase
Fraud alerts at MVP; others in Growth and Scale
Incident Reporting
MVP
Incident Reporting is a structured intake for AI harms. It uses a taxonomy that interoperates with the OECD AI Incidents Monitor and adds community-reported and local-language cases.
Guided report form in English and Urdu with harm type, sector, system and affected group
Six-state evidence model, source trail and confidence label on every record
Right of reply workflow and published corrections history
Encrypted, anonymous whistleblower channel for insiders
Victim-centred harm case desk: evidence preservation, triage and referral to legal aid, platforms or regulators
Judge an AI system, a vendor or a deployment before it reaches people.
AI Risk Assessment
MVP
AI Risk Assessment places a system in its use context and maps it to EU AI Act risk tiers and NIST AI RMF functions. The output is a remediation plan you can act on.
Compliance readiness self-assessment that tiers risk in under an hour
Likelihood and severity scoring with the Global AI Risk Radar as external context
Remediation plans with owners and due dates, exported to the Workbench
Re-assessment triggers when a model, vendor or regulation changes
For
SMEs, IT exporters, risk and compliance teams
Built on
GC-02, RM-01, GC-03
Phase
MVP self-assessment; full scoring in Growth
AI Ethics Evaluation
Growth
AI Ethics Evaluation moves from principle statements to tests. It covers bias, explainability, accessibility and human rights, with an independent ethics board for contested cases.
Community red-team and bias bounty programme with rewards for documented findings
Explainability and contestability review: can affected people understand and challenge a decision?
Accessibility and inclusion audit for low-literacy, low-bandwidth and disabled users
Ethics case library and published advisory opinions
For
AI developers, public bodies, auditors, researchers
Built on
EB-04 to EB-09
Phase
Growth (months 9–18) and Scale
AI Impact Assessment
MVP
AI Impact Assessment is a structured assessment of a system's effects on people, their rights, their jobs and the local economy, carried out before and during deployment.
Algorithmic Impact Assessment (AIA) toolkit with guided templates
Human Impact Statement template: roles affected, support offered, how gains are shared
AI exposure and task-transformation map by occupation
Local economy impact calculator for cities and districts
For
Employers, governments, HR and ESG teams
Built on
EB-03, WF-05, EC-08
Phase
AIA toolkit at MVP; maps and calculator later
AI Vendor Assessment
Growth
AI Vendor Assessment gives buyers a trust profile for every AI platform, model provider and integrator, along with the due-diligence evidence to back a procurement decision.
Vendor directory with trust profiles and a due-diligence document library
Standard procurement questionnaire mapped to your chosen frameworks
AI pricing transparency index and cost and ROI benchmarks
Incident, change-log and transparency score history per vendor
For
Procurement, CIOs, public buyers
Built on
GC-05, EC-02, EC-06, AT-02
Phase
Growth (months 9–18)
Govern
Turn principles and regulation into practice you can audit.
AI Governance Frameworks
Growth
The Standards & Controls Crosswalk maps one set of organisational controls across NIST AI RMF, ISO/IEC 42001, the EU AI Act and the UNESCO Recommendation, so you implement once and evidence many times.
Control crosswalk with gap analysis per framework
Governance starter kit, model policies and playbooks for public and private bodies
Board and executive AI oversight briefings and checklists
Maturity scoring tracked quarter by quarter
For
Boards, CISOs, risk and compliance leads
Built on
GC-09, GC-03, GC-07
Phase
Growth (months 9–18)
AI Policy Management
Growth
AI Policy Management, part of the Governance Workbench, keeps your AI inventory, policies, approvals and attestations in one versioned, auditable place.
AI system inventory with owners, purposes and risk tiers
Policy library with approval workflows, staff attestations and version history
Internal AI-use register you can publish, as we do
Audit-ready exports with full change logs
For
Organisations of every size
Built on
GC-03, GC-07
Phase
Growth; first paid product
Regulatory Compliance
MVP
Regulatory Compliance is a live map of AI laws, guidance and standards. Because deadlines keep moving, it is kept as a change log rather than a fixed timeline.
Global AI Regulation Tracker for the EU, US federal and states, UK, Gulf and Pakistan
Obligation maps by role (provider, deployer, importer) and alerts when they change
Export compliance desk for IT firms and freelancers serving EU, UK, US and Gulf clients
National AI Policy Implementation Monitor that tracks Pakistan's 2025 policy targets
For
Legal, compliance, product teams, IT exporters
Built on
GC-01, GC-06, GC-08, GC-10
Phase
Launch / MVP, months 3–9
AI Compliance Audits
Growth
AI Compliance Audits are fixed-scope, independent reviews of governance maturity, transparency disclosures and bias testing. Each follows a published methodology, and certification decisions are firewalled.
Independent Assessment & Assurance programme with defined scope and fixed fee
Independent bias testing lab for hiring, credit and public-service systems
Accredited auditor network licensed under the platform's methodology
Responsible AI Organisation and Fair AI Transition Employer certifications
For
Banks, employers, public bodies, AI developers
Built on
CT-07, EB-01, CT-06, CT-01, CT-04
Phase
Pilots in Growth; network in Scale
Report & alert
Publish what you know, and hear about changes before they reach you.
AI Transparency Reports
Growth
AI Transparency Reports rate how well AI providers document their systems and give organisations templates to publish their own disclosures.
AI provider transparency scorecards based on documentation quality
Standard disclosure templates covering purpose, training data, limits and human oversight
AI Nutrition Label: a consumer-facing summary of an AI product
Hosted Human Impact Statements with recognition for early adopters
For
AI providers, enterprises, consumers
Built on
AT-02, CT-05, AT-04
Phase
Growth; labels in Scale
AI Accountability Dashboard
MVP
The AI Accountability Dashboard shows what organisations said about AI, what actually happened and how they responded when issues were raised.
Interactive insight dashboards by sector, country and organisation
AI claims tracker that tests "AI-powered" and "AI-driven layoff" claims for AI-washing
Response and remediation rates for named organisations
Corporate AI Accountability Ranking, published only after dashboards are established
For
Investors, ESG analysts, journalists, regulators
Built on
DI-04, AT-06, WF-02, DI-02
Phase
Dashboards at MVP; ranking in Growth
Risk Intelligence Hub
Growth
The Risk Intelligence Hub turns incidents, labour data and regulatory change into early-warning signals for sectors, occupations and regions.
Global AI Risk Radar across technical, legal, human-rights, economic, environmental and geopolitical risk, with severity and time-to-remedy trends
Sector early-warning alerts when indicators cross defined thresholds
Election and information integrity watch
Scenario libraries and stress tests for sector-level AI disruption
For
Policymakers, insurers, investors, civil society
Built on
RM-01, RM-02, RM-04, DI-11
Phase
Radar in Growth; early warning in Scale
Watchlists & Alerts
MVP
Watchlists & Alerts let you follow the vendors, systems, jurisdictions and topics you care about, and get a signal when something material changes.
AI WatchDog Signal: a public alert stream for major incidents and rule changes
Personal and team watchlists delivered by email, in-app or webhook
The annual AI Watch List and AI Good Practice List, each entry with a right of reply
Weekly Watch Bulletin newsletter
For
Everyone; team alerts on paid plans
Built on
RM-09, AT-04, PA-04
Phase
Launch / MVP, months 3–9
Free public programmes
Free for individuals, funded by grants and by the commercial arm. These programmes produce the first-person evidence that every other module relies on.
Displaced & Affected Worker Registry
Private registration with verification tiers, longitudinal follow-up and anonymised dashboards. It opens with the "Count Me In" campaign and a live counter that shows verified and unverified records separately.
WF-01, WF-06
Reskilling, Rights & Talent
Reskilling and redeployment hub, a country-specific worker rights navigator, a freelancer income monitor and a human talent marketplace.
WF-03, WF-07, WF-09, WF-10
Creator Rights & Licensing
Rights registry and consent ledger, "Was My Work Used?" lookup, voice and likeness replica protection, similarity case intake, takedown toolkit, and later a Fair AI Licensing Marketplace.
CR-01 to CR-11
Community, Academy & Research
Multi-track forum, resilience circles, AI Literacy Academy, local chapters, research fellowships and an open data commons.
One workspace for your AI inventory, risk posture, compliance readiness and incident register, built on the same evidence model as the public record. Click through the tabs below. All figures are sample data.
app.aiwatchdog.example/workspace/overviewSample data
Overview
Last 12 months
AI systems inventoried
42
+6 this quarter
High-risk systems
7
No change
Open incidents
11
+3 vs last month
Overall readiness
72%
+9 pts
Risk score trend lower is better
Your riskSector median
Readiness by framework
Compliance scorecards
EU high-risk duties from 2 Dec 2027
Impact assessments due
8
3 this month
Vendor reviews complete
34 / 42
+5 this quarter
Audit packs exported
19
Year to date
Policies awaiting attestation
5
2 overdue
Open obligations by owner
Obligation
Framework
Owner
Due
Status
Human oversight procedure for credit model
EU AI Act
Risk
Q2 2027
In progress
AI literacy training for customer service staff
EU AI Act
People
Done
Complete
Measure function: bias metrics for onboarding chatbot
The same evidence base is configured for each sector's rules, risks and people. Choose an industry.
Government & regulators
Governments are deploying AI in public services and setting national AI policy, often without local evidence on what it changes.
Track public-sector AI deployments and procurement transparency
Monitor national AI policy implementation against stated targets
Use the Workforce Transition Index and local impact calculator for labour policy
Run citizen panels and regulatory sandboxes with documented results
EC-04, GC-08, DI-01, DI-12, GC-04, PA-09
Evidence you can cite
Local, verified data for consultations, budget decisions and donor reporting.
Procurement you can defend
Risk assessments and Transparency Cards for every system before contract award.
Banks, insurers & fintech
Credit scoring is high-risk under the EU AI Act, voice-clone fraud targets customers directly, and investors want comparable AI risk signals.
Independent bias testing of credit and underwriting models
Deepfake and AI fraud alerts for helplines and onboarding
Vendor due diligence for model providers and integrators
Licensed data and index feeds for ESG, investment and insurance analysis
EB-01, RM-03, GC-05, DI-05, CT-07
Fewer surprises
Model drift, policy changes and fraud patterns flagged before customers feel them.
Audit-ready
A technical file and oversight evidence for each high-risk system.
Healthcare & life sciences
Clinical and administrative AI affects patient safety and equity, and errors in regional languages often go unrecorded.
Algorithmic impact assessments before deployment of triage and diagnostic tools
Incident reporting with patient-safety severity levels
Explainability and contestability reviews for decisions about patients
Urdu and regional-language quality benchmarks for patient-facing assistants
EB-03, AT-01, EB-08, EB-02
Safer adoption
Known limits written down and monitored, not discovered in the ward.
Equity in view
Performance gaps by language, region and group surfaced early.
Telecommunications & ISPs
Operators carry the traffic for AI-enabled scams, run AI in customer service and network operations, and employ large contact-centre workforces.
AI fraud and spoofed-call alert sharing with banks and regulators
Governance Workbench for AI in customer care, churn and network automation
Fair AI Transition planning and Human Impact Statements for contact-centre staff
Cybersecurity and supply-chain watch for AI-assisted attacks
RM-03, RM-07, GC-03, WF-08, CT-04
Customer trust
Visible action against AI scams that use your network.
Responsible automation
Redeployment plans published, not announced after the fact.
Education
AI used in admissions and assessment is high-risk, students need AI-resilient career guidance, and educators need practical AI literacy.
AI Literacy Academy courses for staff and students in English and Urdu
Entry-level and youth opportunity watch for graduate hiring trends
Risk assessment for proctoring, grading and admissions tools
Schools and youth AI awareness programme and research fellowships
CE-02, WF-11, GC-02, PA-07, CE-04
Prepared graduates
Curricula guided by real task-transformation data.
Fair assessment
Tools checked for bias and appeal routes before exam season.
Media, publishing & creators
Newsrooms face deepfakes and election misinformation. Publishers and creators need evidence of training use and a route to licensing.
Media desk with verified data and rapid comment on AI stories
Election and information integrity watch with provenance checks
Creator Rights Registry, "Was My Work Used?" lookup and takedown toolkit
Copyright case and settlement tracker, and later a licensing marketplace
PA-05, RM-04, CR-01, CR-02, CR-08, CR-06, CR-04
Faster verification
Synthetic media checked against C2PA content credentials.
Rights with leverage
Pooled evidence for collective negotiation.
Enterprises & IT exporters
Companies serving EU, UK, US and Gulf clients now face AI requirements in contracts. Boards want oversight, and staff want to know what happens to their jobs.
Export compliance desk mapping client-country AI obligations
Independent research approved by the Research & Methodology Council, with published methods and versioned indexes. Non-sensitive data is open for reuse.
Flagship
State of AI Impact 2027
Annual synthesis across every pillar, with a Pakistan and Global South lens.
Workforce
AI and the Future of Work in Pakistan
Task change, displacement claims, reskilling and employer practice.
Economics
State of AI Adoption and Spending
Baseline survey across sectors and organisation sizes.
Risk
AI Incident Trends
Patterns, reporting gaps and lessons by sector.
Creators
Creator Rights in the Generative AI Era
Licensing options, rights preferences and practical barriers.
Language
AI in Urdu and Regional Languages
Quality, access, cultural context and inclusion gaps.
Compliance
EU AI Act after the Digital Omnibus
What moved, what didn't, and what Pakistani IT exporters should do now.
Procurement
AI Procurement Readiness
How organisations evaluate vendors, risk and value.
Regulation
Pakistan National AI Policy: implementation monitor
Progress against the 2025 policy's targets and institutions.
Regulation
US state AI laws: change log
Colorado's replacement act, California SB 53 and federal pre-emption activity.
Economics
Who Captures AI Productivity Gains?
Evidence review of wages, prices, employment and business value.
Public sector
AI in Public Services
A transparency and accountability inventory of selected deployments.
No reports match that search. Try a broader term or clear the filter.
Flagship indexes
Dashboards come first and composite indexes follow, once enough data is in place. Each index has a published methodology, a version history and an independent review.
AI Displacement & Workforce Transition Index
Displacement, transition and recovery by sector, occupation and region, built from registry and panel data.
DI-01
Corporate AI Accountability Ranking
How organisations disclose, govern and remediate AI, with a right of reply before publication.
DI-02
Public AI Trust Barometer
A recurring survey of public awareness, trust, disclosure experience and preferred redress.
DI-03
Glossary
The terms we use, defined once.
AI-washing
Overstating the role of AI, such as blaming layoffs on AI when other factors are the main cause, or marketing ordinary software as AI-powered.
Algorithmic Impact Assessment (AIA)
A structured assessment of an AI system's effects on people and their rights, carried out before and during deployment.
Transparency Card
A standardised, plain-language profile of an AI system: purpose, limits, data, human review, appeal routes and evaluation coverage.
High-risk AI system
Under the EU AI Act, AI used in sensitive areas such as recruitment, credit, education, law enforcement and critical infrastructure.
Right of reply
The opportunity given to a named organisation to respond before or alongside publication.
Memorisation
When an AI model stores parts of its training data, such as text or lyrics, and can reproduce them.
ISO/IEC 42001
The international standard for AI management systems.
C2PA content credentials
An open technical standard for attaching tamper-evident provenance information to digital content.
The Weekly Watch Bulletin
Incidents, rule changes and new research, every week in one email. Free.
The public-interest core is free. Organisations pay for governance tools, assurance and intelligence. No plan, at any price, can influence a ranking, an incident record or a finding.
Public
Workers, creators, consumers and anyone reporting an incident
Illustrative pricing in USD. Purchasing-power-adjusted local pricing is available for Pakistan and other developing markets.
Custom deployment options
For organisations with data residency, sovereignty or integration requirements.
Managed cloud
Multi-tenant SaaS with regional hosting. Fastest to start.
In-country hosting
Single-tenant deployment in your jurisdiction for regulated sectors.
Private cloud or on-premises
Workbench and assessment modules inside your own environment.
National node
A localised observatory for a country or region that feeds the global evidence base.
Research Partner
Controlled-access research datasets, API access, methods workshops and collaborative studies. Available as an institutional membership or project agreement.
Feature comparison across Public, Professional, Organisation and Enterprise plans
Feature
Public
Professional
Organisation
Enterprise
Observe
Watchlist & Transparency Cards
✓
✓
✓
✓
Incident reporting
✓
✓
✓
✓
Model and policy change log
Public feed
✓
✓
✓
Assess & govern
Compliance self-assessment
Basic
✓
✓
✓
Governance Workbench
–
–
✓
✓
Standards & Controls Crosswalk
–
Read-only
✓
✓
Vendor due diligence
–
Profiles
✓
✓
Intelligence
Regulatory alerts
Public signal
✓
✓
✓
Benchmarks
Public
Selected
✓
✓
Data & index API
–
–
–
✓
Security & support
SSO and audit logs
–
–
✓
✓
Data residency options
–
–
–
✓
Support
Community
Email
Priority
Dedicated
Pricing questions
Why is the platform free for individuals?
Affected workers, creators and the public provide the evidence the platform depends on. Charging them would undermine both trust and participation. Grants and organisational subscriptions fund the public core.
Does paying change how we appear in rankings?
No. This is written into the constitution of both entities. Research and editorial staff are firewalled from sales, and certification decisions are made by a separate committee.
Do you offer non-profit or academic pricing?
Yes. Civil society organisations, universities and newsrooms qualify for reduced pricing or a Research Partner agreement.
AI WatchDog grew out of the AI Ethics & Economic Impact Working Group. People who lose work or income to AI have nowhere to be counted, and creators whose work trains AI models have no practical way to know about it or be paid. We are building that place.
Mission and vision
Mission
To make the human, economic and social impact of AI visible, measurable and accountable. We collect reliable evidence, turn it into accessible intelligence, provide practical accountability tools, amplify affected voices, protect creative and human rights, and support responsible AI adoption.
Vision
A world where AI innovation is transparent, accountable, inclusive and beneficial: its benefits shared fairly, its harms seen, counted and corrected, and where individuals and institutions can understand, challenge and govern its effects.
Governance philosophy
Eleven principles, applied to the platform as strictly as to the organisations it watches.
Evidence before accusation
Every claim has a source, a confidence level and a method. Corrections are public.
Independence
No pay-to-rank, no pay-to-remove and no undisclosed conflicts. Funders are disclosed.
Proportionality
Scrutiny matches potential impact, so powerful actors and high-stakes systems get the closest look.
Privacy and dignity
Affected people control their data. We publish only aggregates and tell stories only with consent.
Constructive, not anti-AI
We recognise good practice and positive transitions as well as harms.
Right of reply and appeal
Named organisations are notified, and their responses are published alongside our findings.
Explainability
Every rating and automated step can be explained to the people it affects.
Openness
Methods, taxonomies and non-sensitive data are open for scrutiny and reuse.
Inclusion
Urdu and regional languages, accessibility, low-bandwidth design and gender balance.
Practise what we preach
We publish our own AI-use register and hold ourselves to the same standards.
Environmental sustainability
We use compute responsibly and treat environmental cost as part of AI accountability.
Two entities, one firewall
The watchdog and the business are separate by design.
AI WatchDog Foundation
A not-for-profit company. It runs the registry, incident database, research, indexes, Watch List, community, advocacy and open data, and owns the brand.
Independence firewall
AI WatchDog Services
A commercial company owned by the Foundation. It provides SaaS, training, audits, data licensing and events, has a separate P&L and has no influence over public findings.
Leadership team
Placeholder profiles. Add names, photos and bios once appointments are confirmed.
[Name]
Executive Director
[Name]
Head of Research & Methodology
[Name]
Head of Product
[Name]
Editor & Investigations Lead
Advisory board and governance bodies
Multi-stakeholder by design. Each body has a published mandate, a register of interests and term limits.
Board of Trustees
7 to 11 members from civil society, academia, law, labour, creators, technology and finance. A majority are independent, and no sponsor has a veto.
Research & Methodology Council
Approves research design, indexes and Watch List entries, and ensures peer review and reproducibility.
Ethics, Rights & Safety Committee
Reviews sensitive investigations, vulnerable participants and responsible disclosure.
Community Council
Elected by verified registrants and creators. It advises on priorities, moderation and campaigns.
AI Ethics & Economic Impact Working Group
The founding group. It incubates the platform and later becomes the Advisory Council.
Advisory Council
Industry, government and international organisations. It is advisory only and has no say over findings.
Roadmap
We move to each phase when its milestones are met, not when a date arrives.
Months 0–3Foundation
Working Group charter, legal entities, methodology v1, demand survey, pre-registration.
Months 3–9Launch / MVP
21 critical features: registry, trackers, Watchlist, community, first report.
Months 9–18Growth
First indexes, assessment and certification pilots, Workbench revenue.
Months 18–36Scale
South Asia and Gulf coverage, licensing marketplace, data API, early warning.
Year 3–5+Institutional & global
Global South coalition, recognised standards, formal policy roles.
Choose the right channelCount Me In for work and income changes, Incident Reporting for harms caused by an AI system, or the Creator Rights Registry for your works.
Gather what you haveDates, the organisation or product involved, and any letters, screenshots or contracts. A report without documents is still welcome and is recorded as self-reported.
Decide what you shareYou choose which details are stored, whether your story can be quoted, and whether we may contact you again. You can change or withdraw consent later.
Submit in English or UrduThe guided form asks one thing at a time. Most people finish in under ten minutes.
Review and evidence gradingA reviewer assigns an evidence state. Named organisations are given a right of reply before anything about them is published.
Follow-up and referralRegistry participants get check-ins at 3, 6 and 12 months. Where useful, we refer you to legal aid, reskilling or the relevant platform or regulator.
Evidence & publication standards
Every record carries one of the six evidence states shown on the home page. These rules decide what can be published and how individuals are protected.
Where records at each evidence state can appear
Evidence state
Individual record
Aggregate statistics
Public alerts and rankings
Self-reported
As testimony, only with consent
Yes, labelled as unverified
No
Document-supported
Yes, with evidence type shown
Yes
Only with corroboration
Corroborated
Yes
Yes, with caveats
Yes, after human review
Independently verified
Yes, with review scope and date
Yes
Yes
Disputed or under review
Neutral status only
Excluded until resolved
No
Corrected or withdrawn
Correction history kept where lawful
Recalculated
Removed or corrected
Publication safeguards
Identity data is stored separately from impact data
Minimum cohort sizes and privacy-preserving methods for every published aggregate
Human review before any public alert or sensitive submission is published
No causal claim, such as "AI caused this job loss", without supporting evidence
Encryption, access controls, audit logs and defined retention periods
A documented security incident response and a responsible disclosure route for researchers
More questions
The basics, such as what AI WatchDog is and how ranking independence works, are on the home page.
Using the platform
Who can use AI WatchDog?
Anyone. Workers, freelancers, creators, consumers, researchers, journalists, civil society, policymakers, businesses and AI developers can all use the public resources. Organisations add paid tools on top.
Where does it operate first?
The pilot focuses on Pakistan, especially Lahore and Islamabad, along with Gulf-linked labour markets and South Asia. Wider Global South and international coverage follows.
How do I report an AI-related harm?
Use Incident Reporting. The harm case desk takes you through a guided intake, preserves your evidence, triages the case and, where useful, refers you to legal aid, the platform involved or a regulator. See the reporting guide above.
Privacy & creators
How is my personal information protected?
We apply privacy by design: data minimisation, purpose limitation and consent you control. Identity data is kept separate from impact data, everything is encrypted with clear retention rules, and aggregates use minimum cohort sizes and privacy-preserving methods. Individual reports are never public by default.
What is the Creator Rights Registry?
A neutral rights and provenance record where creators register works and state their AI training preferences, such as opt-out or licence. It does not assume that every training use is unlawful or that payment is automatic. Licensing pilots come later.
How do I request a correction or right of reply?
Choose "Report a correction" on the contact page and name the record. Named organisations are notified before publication and can respond at any time afterwards. Responses are published alongside the finding.
For organisations
How do organisations use the Governance Workbench?
It brings together your AI inventory, risk classification, impact assessments, vendor due diligence, policy approvals, monitoring and exportable audit packs, for regulatory readiness and internal accountability.
Is the platform legal or compliance advice?
No. The regulation tracker and readiness tools give information and structured self-assessment. They don't replace qualified legal advice for your jurisdiction.
When will certification be available?
Only after the methodology, assessor independence, appeals process and audit controls are mature. Early work focuses on evidence, governance tools and fixed-scope independent assessments.
Where is the methodology published?
Every index, ranking and assessment has a published methodology covering dimensions, weights, data coverage, uncertainty and version history, reviewed by the Research & Methodology Council.
No questions match that search. Try another word or ask us directly.