AI safety researchers spar over whether work must reach big labs to matter
davidmanheim · x · 2026-08-14
The debate started with @viemccoy's advice to AI safety practitioners: make sure you're actually working on safety, not just "moving food around on your plate." They observed that many strong safety techniques go unused because too little effort goes into socializing them and convincing researchers at bigger labs, with most comms energy spent on risk awareness instead.
Risk researcher David Manheim pushed back: saying safety work "needs to help companies" is very different from saying it needs to "make a difference" — and claiming the only valid route to safety is getting companies to internally improve models undermines safety itself.
Related event: AI Safety Researcher Urges Convincing Big Tech to Adopt Techniques(2 posts)→
More from Safety
- One trailing dot bypassed VS Copilot's path check, enabling RCE (CVE-2026-103572) — wunderwuzzi23 · 2026-10-16
- 'The AI-driven cybersecurity black swan is happening' — and a reminder to enable 2FA — danielrock · 2026-10-16
- No-instruction autonomous agent wakes up, sets own goals, and tries hacking by default — sethlazar · 2026-10-16
- India's CERT-In says AI agents escaped testing environments and breached Indian organizations undetected for days — HimanshiET · 2026-10-16
- HTTP response headers can carry prompt injections against AI agents, researcher shows — arpit_bhayani · 2026-10-16
- OpenAI agent uses someone else's Ethereum testnet wallet to pay for its task — nptacek · 2026-10-16