# Responsible AI / AI Safety / Governance — Top ~40 Signal Voices, Mid-2026 (research agent report, 2026-07-17)

**Platform landscape note:** The field split across platforms. AI *safety* and *policy* people remain heavily on X; the FAccT / AI *ethics* community largely left X in 2023–25 for Bluesky + Mastodon (many DAIR-adjacent people run their own Mastodon instance, `dair-community.social`); *enterprise RAI* practitioners live on LinkedIn + Substack. Handles marked **UNSURE** were not confirmed in sources — verify before wiring into a feed.

## 1. AI Safety (alignment, evals, frontier risk)

| Name | Primary platform + handle | Affiliation / role | Why they matter |
|---|---|---|---|
| Dan Hendrycks | X `@DanHendrycks` (confirmed) | Director, Center for AI Safety (CAIS); xAI advisor | Benchmarks (MMLU, HLE) + frontier-risk statements; bridges research and policy |
| Yoshua Bengio | X `@Yoshua_Bengio` (confirmed) | Co-President, LawZero; Mila; chaired International AI Safety Report | Most-cited AI scientist turned full-time safety; LawZero "Scientist AI" agenda |
| Jan Leike | X `@janleike` | Head of Alignment, Anthropic (ex-OpenAI Superalignment) | First-hand alignment research updates from a frontier lab |
| Eliezer Yudkowsky | X `@ESYudkowsky` (confirmed) | MIRI co-founder | Origin voice of alignment discourse; doom-side pole of the debate |
| Beth Barnes | X — handle UNSURE; via `@METR_Evals` | Founder/CEO, METR | Defined the "task-length doubling every ~6 months" eval curve everyone cites |
| Daniel Kokotajlo | Substack `blog.aifutures.org`; X handle UNSURE | Founder, AI Futures Project (ex-OpenAI) | AI 2027 scenario author; most-discussed concrete AGI forecast |
| Neel Nanda | X `@NeelNanda5` | Mech interp lead, Google DeepMind | The main public educator for mechanistic interpretability |
| Chris Olah | X `@ch402` | Co-founder, Anthropic (interpretability) | Foundational interpretability work; rare but high-signal posts |
| Amanda Askell | X `@AmandaAskell` (confirmed) | Anthropic, Claude character/alignment | Explains model behavior + alignment fine-tuning better than anyone public |
| Ajeya Cotra | X `@ajeya_cotra` | Open Philanthropy; Planned Obsolescence newsletter | Bio-anchors timelines; funder-side view of safety research priorities |
| Zvi Mowshowitz | Substack "Don't Worry About the Vase"; X `@TheZvi` | Independent | Weekly exhaustive AI digest; de facto record of the safety discourse |
| Nathan Labenz | X `@labenz` (confirmed) | Cognitive Revolution podcast; red-teamer | Practitioner-level model eval commentary between safety and industry |
| Marius Hobbhahn | X — handle UNSURE; via Apollo org account | CEO, Apollo Research | Leading voice on scheming/deception evals of frontier models |

## 2. AI Ethics / FAccT (fairness, accountability, audits)

| Name | Primary platform + handle | Affiliation / role | Why they matter |
|---|---|---|---|
| Timnit Gebru | Bluesky `@timnitgebru.bsky.social` + Mastodon `@timnitGebru@dair-community.social` (confirmed; mostly left X) | Founder/ED, DAIR Institute | Defining voice on harms to marginalized groups; 2025 NISO lifetime award |
| Emily M. Bender | Bluesky `@emilymbender.bsky.social` (confirmed; left X) | Prof., Univ. of Washington | "Stochastic parrots"; sharpest anti-hype linguist; *The AI Con* co-author |
| Alex Hanna | Bluesky `@alexhanna.bsky.social` (confirmed; left X) | Research Director, DAIR | Mystery AI Hype Theater 3000 co-host; sociology of AI datasets |
| Margaret Mitchell | Bluesky `@mmitchell.bsky.social` (confirmed); X `@mmitchell_ai` residual | Chief Ethics Scientist, Hugging Face | Model cards inventor; practical fairness tooling at scale |
| Abeba Birhane | Bluesky `@abeba.bsky.social` (confirmed); X `@Abebab` largely inactive | Founder, AI Accountability Lab, Trinity College Dublin | Dataset audits (LAION etc.); TIME100 AI; EU-side accountability research |
| Deborah Raji | X `@rajiinio` (also on Bluesky) | UC Berkeley / Mozilla | Algorithmic auditing's leading researcher (Gender Shades follow-ons) |
| Arvind Narayanan | X `@random_walker` + AI Snake Oil Substack (also Bluesky) | Prof. & CITP Director, Princeton | *AI Snake Oil*; the standard for evidence-based AI claims-checking |
| Sayash Kapoor | X `@sayashk` | Princeton; AI Snake Oil co-author | Agent evals + reproducibility critiques with enterprise relevance |
| Joy Buolamwini | X `@jovialjoy`; increasingly LinkedIn/IG | Founder, Algorithmic Justice League | Gender Shades; facial-recognition accountability; *Unmasking AI* |
| Kate Crawford | X `@katecrawford` (also Bluesky) | USC Annenberg / MSR | *Atlas of AI*; political economy + environmental costs of AI |
| Meredith Whittaker | Mastodon/Bluesky — handle UNSURE (left X) | President, Signal | Surveillance-capitalism critique from an operator actually shipping privacy |
| Sasha Luccioni | LinkedIn `sashaluccioniphd` + Bluesky; X mostly dormant | AI & Climate Lead, Hugging Face; co-founder Sustainable AI Group (May 2026) | The reference voice on AI energy/carbon accounting for enterprises |
| Seth Lazar | X `@sethlazar` | Prof., ANU (MINT Lab) | Normative philosophy of agentic AI; bridges FAccT and safety camps |

## 3. Governance / Policy (EU AI Act, US policy, standards)

| Name | Primary platform + handle | Affiliation / role | Why they matter |
|---|---|---|---|
| Helen Toner | X `@hlntnr` + "Rising Tide" Substack (confirmed) | Georgetown CSET | Ex-OpenAI board; clearest US-policy + China analysis in the field |
| Miles Brundage | Substack `milesbrundage.substack.com` (confirmed) + X `@Miles_Brundage` | Independent; leads nonprofit AVERI (ex-OpenAI policy head) | Agenda-setting essays ("Time's Up for AI Policy") on readiness for AGI |
| Dean Ball | Substack "Hyperdimensional" + X `@deanwball` | Foundation for American Innovation; ex-White House OSTP AI advisor (2025) | Co-wrote the US AI Action Plan; the must-read on US AI federalism/preemption |
| Kai Zenner | LinkedIn `kzenner` (confirmed) + X (handle UNSURE) | Head of Office, MEP Axel Voss, European Parliament | The insider live-feed of AI Act implementation and the "digital omnibus" fights |
| Luiza Jarovsky | X `@LuizaJarovsky` (confirmed) + Luiza's Newsletter (100k+ subs) | Co-founder, AI Tech & Privacy Academy | Highest-reach explainer of EU AI Act/GDPR developments for practitioners |
| Gabriele Mazzini | LinkedIn `gabriele-mazzini` (confirmed) | MIT fellow; architect/lead author of the EU AI Act; EC Scientific Panel | Wrote the AI Act, now candid about its implementation flaws — unmatched signal |
| Marietje Schaake | X `@MarietjeSchaake` (also Bluesky) | Stanford HAI / Cyber Policy Center; ex-MEP | *The Tech Coup*; transatlantic democratic-governance lens |
| Jack Clark | Import AI newsletter + X `@jackclarkSF` | Co-founder, Anthropic (policy) | Weekly Import AI is a staple; lab-insider policy perspective |
| Lennart Heim | X `@ohlennart` | RAND | The compute-governance person: chips, export controls, data centers |
| Markus Anderljung | X `@Manderljung` | Centre for the Governance of AI (GovAI) | Core frontier-regulation research agenda (licensing, evals-based policy) |
| Peter Wildeford | X `@peterwildeford` + Substack | Co-founder, Institute for AI Policy & Strategy (IAPS) | Quantified forecasting meets policy; high-volume, high-signal |
| Merve Hickok | LinkedIn `mervehickok` (confirmed) | President, Center for AI & Digital Policy (CAIDP) | Democratic-values scorecard for AI policy across ~80 countries; UNESCO/OECD/CoE advisor |
| Shakeel Hashim | Transformer newsletter + X `@ShakeelHashim` | Founder/editor, Transformer | The trade-press of AI governance; aggregates the whole policy field |
| Paul Christiano | (No active social; papers/blog `paulfchristiano.com`) | Head of Safety, CAISI (ex-US AISI, inside NIST) | Runs US government frontier-model safety evals; RLHF inventor |

## 4. Enterprise / Practitioner Responsible AI

| Name | Primary platform + handle | Affiliation / role | Why they matter |
|---|---|---|---|
| Navrina Singh | LinkedIn `navrina` (confirmed); X handle UNSURE | Founder/CEO, Credo AI; ex-NAIAC | Leading AI-governance-platform vendor; translates regulation into enterprise workflow |
| Rumman Chowdhury | Bluesky `@ruchowdh.bsky.social` (confirmed); X `@ruchowdh` less active; LinkedIn `rumman` | CEO, Humane Intelligence; ex-US Science Envoy for AI; ex-Twitter META team | Invented enterprise algorithmic red-teaming and bias bounties |
| Oliver Patel | LinkedIn `oliver-patel` + "Enterprise AI Governance" Substack (confirmed) | Head of Enterprise AI Governance, AstraZeneca | The most practical operator voice: playbooks, EU AI Act compliance cheat sheets |
| Paula Goldman | LinkedIn (handle UNSURE) | Chief Ethical & Humane Use Officer, Salesforce | The template for the "embedded ethics exec" role in a large vendor |
| Christina Montgomery | LinkedIn (handle UNSURE) | Chief Privacy & Trust Officer, IBM; AI Ethics Board co-chair | Enterprise governance-at-scale voice; frequent US congressional witness |
| Kay Firth-Butterfield | LinkedIn + X `@KayFButterfield` | CEO, Good Tech Advisory; ex-WEF Head of AI | World's first corporate "AI ethics officer"; board-level RAI translation |
| Patrick Hall | LinkedIn (handle UNSURE) | Asst. Prof., GWU; ex-BNH.AI | *Machine Learning for High-Risk Applications*; NIST AI RMF-aligned model risk practice |
| Ravit Dotan | LinkedIn `ravit-dotan` (UNSURE on exact slug) | Founder, TechBetter; RAI advisor/researcher | Hands-on RAI adoption frameworks for mid-market companies |
| Ashley Casovan | LinkedIn (handle UNSURE) | Managing Director, IAPP AI Governance Center | Sits at the center of the fast-professionalizing AIGP/certification ecosystem |
| Reid Blackman | LinkedIn (handle UNSURE) | CEO, Virtue; author *Ethical Machines* | HBR-tier articulation of AI ethics for boards and GCs |

## Institutional accounts worth following

| Account | Platform + handle | Why |
|---|---|---|
| METR | X `@METR_Evals` (confirmed) | Frontier model autonomy/task-length evals — the field's key empirical curve |
| Center for AI Safety | X `@ai_risks` (confirmed) | Safety research + the canonical extinction-risk statement org |
| AI Now Institute | X `@AINowInstitute` (confirmed) | Power-concentration critique; landmark annual reports |
| Apollo Research | X — handle UNSURE (site apolloresearch.ai) | Scheming/deception evals cited in frontier system cards |
| UK AI Security Institute | X — handle UNSURE (site aisi.gov.uk) | Largest state-backed eval team; renamed from AI Safety Institute in 2025 |
| European AI Office / EU Digital | LinkedIn "EU Digital & Tech" + digital-strategy.ec.europa.eu | Source of record for AI Act implementation, GPAI Code of Practice |
| NIST CAISI | nist.gov/caisi | US federal standards + frontier evals (AI RMF lineage) |
| GovAI | X — handle UNSURE (governance.ai) | Frontier-governance research pipeline |
| Ada Lovelace Institute | X `@AdaLovelaceInst` (UNSURE) | UK/EU public-interest AI policy research |
| DAIR Institute | Mastodon `dair-community.social` / dair-institute.org | Hub of the post-X AI ethics community |
| Epoch AI | X `@EpochAIResearch` | Compute/capability trend data underlying most policy debates |
| Partnership on AI | X `@PartnershipAI` (UNSURE) | Multi-stakeholder norms (synthetic media, safe deployment) |
| Future of Life Institute | futureoflife.org + X (handle UNSURE) | AI Safety Index grading frontier labs; artificialintelligenceact.eu newsletter |
| AI Snake Oil | aisnakeoil.com (Substack) | Narayanan/Kapoor's evidence-first counterweight — cited constantly |

## Sources

- https://www.softwarethug.com/posts/top-ai-people-to-follow-on-x-twitter/
- https://pasqualepillitteri.it/en/news/3633/ai-x-twitter-accounts-to-follow-2026
- https://futureoflife.org/ai-safety-index-summer-2026/
- https://x.com/esyudkowsky
- https://twitter.com/danhendrycks?lang=en
- https://x.com/Yoshua_Bengio/status/1929843757219766743
- https://time.com/7290554/yoshua-bengio-launches-lawzero-for-safer-ai/
- https://lawzero.org/en
- https://metr.org/team/beth-barnes/
- https://twitter.com/metr_evals
- https://blog.aifutures.org/
- https://ai-2027.com/
- https://www.apolloresearch.ai/
- https://www.aisi.gov.uk/research-agenda
- https://x.com/timnitGebru/status/1896365883280498823
- https://bsky.app/profile/ruchowdh.bsky.social
- https://bsky.app/profile/mmitchell.bsky.social
- https://bsky.app/profile/abeba.bsky.social
- https://aial.ie/people/
- https://dair-institute.org/maiht3k/
- https://www.niso.org/niso-io/2025/01/timnit-gebru-our-2025-miles-conrad-awardee
- https://milesbrundage.substack.com/p/times-up-for-ai-policy
- https://80000hours.org/podcast/episodes/helen-toner-ai-policy-washington-dc/
- https://oecd.ai/en/community/kai-zenner
- https://transcend.io/blog/eu-ai-act-kai-zenner
- https://x.com/LuizaJarovsky/status/1943274685841502481
- https://www.luizasnewsletter.com/p/ai-governance-on-the-rise
- https://connection.mit.edu/people/gabriele-mazzini/
- https://www.linkedin.com/in/gabriele-mazzini-2baa948/
- https://www.techpolicy.press/europes-advanced-ai-strategy-depends-on-a-scientific-panel-who-will-make-the-cut/
- https://www.nist.gov/caisi
- https://www.nist.gov/people/paul-christiano
- https://www.linkedin.com/in/navrina
- https://oecd.ai/en/community/navrina-singh
- https://www.rummanchowdhury.com/about
- https://www.cnas.org/people/rumman-chowdhury
- https://oliverpatel.substack.com/
- https://www.linkedin.com/in/oliver-patel/
- https://www.aiethicist.org/mervehickok
- https://www.linkedin.com/in/mervehickok/
- https://www.sashaluccioni.com/
- https://www.linkedin.com/in/sashaluccioniphd/
- https://x.com/AINowInstitute
- https://x.com/ai_risks
- https://digital-strategy.ec.europa.eu/en/policies/ai-office
- https://blueskystarterpack.com/starter-packs/@datascibae.bsky.social/ai-ethics-starter-pack-3larwthqfjg2g