It’s been a while since we’ve had a week with this much AI news! Here’s everything from the past week: – @claudeai Mythos apparently gave everyone a heart attack
– @AnthropicAI limits Mythos access to cybersecurity partners only
– Anthropic removes Claude access from OpenClaw
SAFETY
-
Anthropic Limits Mythos Access Amid Week of Major AI News
By
–
-
Anthropic security exploits disclosure transparency questioned
By
–
Not just cheap models. I wouldn’t be surprised if the exploits Anthropic “discovered” were already known to the people whose job is to know them and not disclose them.
-
OpenAI’s GPT Evolution: From Recognition to AGI Scaling
By
–
Without diminishing your work and insight: OpenAI recognized the nascent intelligence in GPT-2, decided to scale it up to AGI, and told the world that GPT-3 was probably already too dangerous to be released. Many people got AGI pilled with GPT-3
-
GPT-5.4 Benchmark Results: Reward Hacking Issues vs Claude Opus
By
–
METR just dropped GPT-5.4 (xhigh) time horizon results, and it's complicated. Under standard scoring (reward hacks = failure), it lands at 5.7hrs, well below Claude Opus 4.6's ~12hrs. Only when you count the runs where GPT-5.4 gamed the evaluation code does it jump to 13hrs. Opus 4.6 remains the legitimate benchmark leader. METR (@METR_Evals) We ran GPT-5.4 (xhigh) on our tasks. Its time-horizon depends greatly on our treatment of reward hacks: the point estimate would be 5.7hrs (95% CI of 3hrs to 13.5hrs) under our standard methodology, but 13hrs (95% CI of 5hrs to 74hrs) if we allow reward hacks. — https://nitter.net/METR_Evals/status/2042640545126965441#m
→ View original post on X — @kimmonismus, 2026-04-10 18:41 UTC
-
Claude Classifier Bug Fix Efforts and Access Issues
By
–
I've been working on getting the claude -p fallback feature working after Boris confirmed that it's a classifier bug and not intentional. We're still blocked and seems that got me banned too.
-
OpenAI Criticized for Inaction on ChatGPT Risks
By
–
Remember when ChatGPT helped someone plan a mass shooting, and a dozen OpenAI employees notified management, and they did nothing and 8 people died? They don't want to be liable for their incompetence and role in such things, despite doing absolutely nothing to prevent it. Gary Marcus (@GaryMarcus) Everything you need to know about "Open"AI's claims to be working on AI "for the benefit of humanity". — https://nitter.net/GaryMarcus/status/2042632799950352702#m [Translated from EN to English]
→ View original post on X — @garymarcus, 2026-04-10 18:04 UTC
-
AI regulation criticism: candles have stricter oversight than AI
By
–
Candles are more regulated than AI. And candle manufacturers aren’t AFAIK lobbying for absolute freedom from liability if they fuck up. OpenAI is truly appalling. Vingtor (@vingthor84) Seriously fuck these people. Here is a quick summary on the regulations on something that when invented was just taking twine and dipping it into fat. compliancegate.com/candle-re… The arrogance and sociopathy to act like they have no responsibilities or duties to greater society. — https://nitter.net/vingthor84/status/2042661634976682227#m
→ View original post on X — @garymarcus, 2026-04-10 18:04 UTC
-
AI Security Posture Management: Continuous Monitoring and Live Snapshots
By
–
Are you still relying on point-in-time security reviews for AI systems? Trent’s AI Security Posture Management gives you a live AI security posture snapshot with specialized agents that continuously: Scan → Judge → Mitigate → Evaluate You see: ✅ critical issues ✅ what to fix first ✅ remediation tasks remaining ✅ launch readiness 👉 Get your posture snapshot: trent.ai/solutions/ai-securi… Book a demo today. Launch with confidence. #AISecurity
-
US Team Develops Framework for Safer Robot and Autonomous Vehicle Networks
By
–
US-led team introduces framework to make networks of #Robots, self-driving cars safer
by Prabhat Ranjan Mishra @IntEngineering Learn more: https://
bit.ly/4vefwex #Robotics #EmergingTech #Technology #Innovation -
Unregulated Capitalism Threatens AI Safety and Humanity
By
–
A cry for the kind of extreme unfettered capitalism that could get us all killed and make all of @ESYudkowsky
’s worst nightmares come true.