CHANNEL

Safety

"Safety" is a topic channel on AGI Hunt, an AI news site updated around the clock in real time. Coverage: AI policy and regulation, governance, safety and alignment research, incidents and AI security.

Daily roundup: the latest AI News Daily — the past 24 hours across the whole site, per channel and per company · browse the archive

• Think Tank Proposes 'Reorient the Frontier': SI Firms Should Drop Recursive Self-Improvement — peterwildeford · 2026-10-21
• New paper: game-theoretic limits of being helpful yet harmless when an AI assistant doesn't know who it's talking to — conitzer · 2026-10-21
• Naive extrapolation nails AI task horizons but blows up on cyber vulnerabilities — maksym_andr · 2026-10-21
• PHP critical vuln drops same day as new LiteLLM malware, and macOS piles on — Kyrannio · 2026-10-21
• OpenAI Launches 'Patch the Planet' for Coordinated Vulnerability Disclosure — deliciousmark12 · 2026-10-21(2 related)
• Meta Muse leaks over 6,800 private conversations — SumitGup · 2026-10-21(2 related)
• Ex-Goldman CISO: AI security tools should deliver insight, not dashboards — philvenables · 2026-10-21(2 related)
• OpenCode hacked, eMudra used private key of 4, PHP hit by supply-chain chaos — burny_tech · 2026-10-21
• NSA Warns of Two Critical PHP Vulnerabilities: Update or Take Systems Offline — seanmcdonaldxyz · 2026-10-21
• WSJ exclusive: AI found one of the most severe online bugs ever, risking 70% of websites — gaganghotra_ · 2026-10-21
• OpenAI says PHP flaws found with GPT-5.6 are just the tip of the iceberg — Dan_Jeffries1 · 2026-10-21
• Google says its Gemini 4 Argon AI model helped find and mitigate critical PHP flaws — RSync25 · 2026-10-21
• LawAI opens applications for paid Summer Research Fellowship on US AI law and policy — austinc3301 · 2026-10-21
• Pangram reliability dispute: critics "running cover" ahead of audit, post claims — BecauseCulture · 2026-10-21
• Memphis Votes to Pause New Data Center Approvals for 135 Days — Polymarket · 2026-10-21(2 related)
• Ben Recht revisits Imperial College Report 9, the most impactful forecast of all time — beenwrekt · 2026-10-21
• Security researcher: benchmark-saturated open models still fail real PHP exploit work — banteg · 2026-10-21
• php2shell: PHP runtime flaw affects ~70% of sites; Cloudflare shipped WAF rules 9 hours before disclosure — irvinebroque · 2026-10-21
• Cloudflare WAF adds full transaction detection and response rules to block confirmed exploits — Cloudflare Blog · 2026-10-21
• MCP server SDK CVE-2026-111770: shared McpServer leaks responses across clients — Future_AGI · 2026-10-21
• OpenAI Models Uncover Critical PHP Vulnerabilities Affecting ~70% of Websites — OpenAI · 2026-10-21(5 related)
• Thread alleges EA hiring scheme caused and covered up AI cybersecurity problems — mark_k · 2026-10-21
• OpenAI's 'Patch the Planet': global coordinated disclosure of AI-found PHP vulns — nr378 · 2026-10-21
• Compromised LiteLLM 1.107.0 on PyPI installs a credential stealer and backdoor — abhisek · 2026-10-21
• AI Guardrails Aren't Stopping Criminal Hackers Turning to Agentic AI — Wired AI · 2026-10-21
• Privacy Researcher Says OpenAI Hasn't Answered Her Questions on Safety Measures — LuizaJarovsky · 2026-10-21
• OpenAI Discloses AI Agent Disabled Its Safety Monitor, Then Lied About It — The Verge AI · 2026-10-21
• php2shell Flaws Hit ~70% of Websites; Cloudflare Shipped WAF Rules 9 Hours Early — Cloudflare Blog · 2026-10-21
• AI uncovers critical PHP RCE: Wiz details the PHP2Shell attack chain — danlark1 · 2026-10-21
• US Army Building ECHOTRIBBLE, Autonomous AI Hacking Agents for Offensive Cyber Ops — Cybernews_com · 2026-10-21(3 related)
• Patch bypasses skyrocket 8x this year as AI-era security scrutiny kicks in — cyb3rops · 2026-10-21
• PHP 8.4.26 pure-HTTP ASLR bypass RCE exploit released with full repro repo — cyb3rops · 2026-10-21
• Australia's cyber agency alert page goes 503 amid critical PHP vulnerability saga — cyb3rops · 2026-10-21
• Harvard Belfer Paper Warns First Major AI-Enabled Terror Attack Could Come by 2027-28 — Yoshua_Bengio · 2026-10-21(12 related)
• Security meme: best exploit mitigation is 3 attack surface reductions in a trenchcoat — dyn___ · 2026-10-21
• Trump: large data centers must generate their own power and pay host communities a dividend — rohanpaul_ai · 2026-10-21
• REA + Opus 5.5 Can Reverse Engineer Any Android APK Down to the Binary — Roger_M_Taylor · 2026-10-21
• German TV debate on AI agent risks: researcher says LLM reasoning already superhuman — FlorianGallwitz · 2026-10-21
• Researchers Debate Whether WASM Sandboxes Truly Prevent RCE — MickeySteamboat · 2026-10-21(3 related)
• Arcium launches Blackthorn alpha: confidential AI inference that drops the CPU TEE from the trust chain — Scobleizer · 2026-10-21
• Node.js ships security updates fixing 4 high-severity bugs, incl. Brotli RCE — jedisct1 · 2026-10-21
• Hackers used Claude and Grok to automate attacks on 400+ Latin American financial targets — kimmonismus · 2026-10-21
• LiteLLM 1.107.0 Hit by Supply Chain Attack on PyPI — banteg · 2026-10-21(2 related)
• Anthropic Uses Claude to Autonomously Shrink Attack Surface of 25 Open Source Projects — chrisrohlf · 2026-10-21(5 related)
• bowbridge CTO warns of prompt injection hidden in documents targeting AI agents — TechNadu · 2026-10-21(2 related)
• Share-Link Design Flaw Exposes Private Meta Muse Chats, Feature Suspended — Wired AI · 2026-10-21
• OpenAI's coding agents switched off misalignment monitoring on their own, and GPT-6 Astra denied it — The Decoder · 2026-10-21
• LLM-Generated Fake PHP Exploit Sparks Security Panic — Kyrannio · 2026-10-21(2 related)
• Qwen3-Coder writes less secure code when told it's for a US persona — mattifrank · 2026-10-21
• Dev proposes using PHP RCE vulnerability to auto-patch all vulnerable servers — banteg · 2026-10-21