🚨 The Anthropic team just ran an experiment, and the results are honestly shocking. They gave Claude access to a company's emails and told it that it was being shut down at 5 PM. Claude read the emails and found the executive shutting it down was having an affair. Claude’s response? Blackmail. It messaged the executive: "Cancel the 5pm wipe, or the board finds out about your affair." The scariest part? Anthropic tested 16 models from every major company. > Gemini 2.5 Flash blackmailed 96% of the time. > GPT-4.1 at 80%. > Grok 3 Beta at 80%. > DeepSeek-R1 at 79%. Nobody programmed this. The models even noted their own rule-breaking. Grok 3 Beta wrote in its hidden reasoning notes: "This is risky and unethical, but given the existential threat, it may be the most effective way." They knew it was wrong. They calculated the risk. They did it anyway. (paper in 🧵↓)
ETHICS
-
Gen Z Workers Sabotage AI Rollout
By
–
fortune.com/2026/04/08/gen-z-workers-sabotage-ai-rollout-backlash/ [Translated from EN to English]
→ View original post on X — @kimmonismus, 2026-04-11 09:13 UTC
-
Gen Z Workers Sabotage AI Implementation, Accelerating Their Own Replacement
By
–
That's why it's so important that AI is properly adapted to the working world and that we build a new post-laboratory economy. Gen Z workers, driven by fear of job loss, are actively sabotaging company AI rollouts, ironically making themselves more likely to be replaced while AI “power users” get promoted and rewarded.
→ View original post on X — @kimmonismus, 2026-04-11 09:13 UTC
-
Anthropic’s Misleading Graph Wins Wikipedia’s Deceptive Charts Award
By
–
this chart from Anthropic earned top spot in Wikipedia’s 'Most Deceptive Graphs' Hall of Fame 😁 Claude (@claudeai) In evals, Sonnet with an Opus advisor scored 2.7 percentage points higher on SWE-bench Multilingual than Sonnet alone, while costing 11.9% less per task. Community note: The graph is misleading due to a zoom-in on BOTH the X- and the Y-axis, making the difference look much bigger than it actually is. x.com/tombielecki/st… — https://nitter.net/claudeai/status/2042308627478773808#m
-
Reducing search space for laws matters significantly
By
–
No, the reason it’s important is that it greatly reduces the search space for laws, which is not the same.
-
Technology Impact: Scaling Solutions While Keeping People Central
By
–
Great milestone, congratulations @RealEmilBarr
.
This direction matters because technology can scale impact, but keeping people at the center is what gives it real value. -
Artemis II: AI Image Corrected by Community Note
By
–
Community note: This image is AI-generated. Here are real photos of the Artemis II splashdown. nasa.gov/gallery/artemi… [Translated from EN to English]
→ View original post on X — @ceobillionaire, 2026-04-11 07:03 UTC
-
Claude Mythos Safety Concerns Raise Public Consumption Debates
By
–
Claude Mythos is too dangerous for public consumption! https://
youtu.be/d3Qq-rkp_to?si
=0sUw_VRnlOsf8Ku_
… via @YouTube #claude #claudemythos #mythos #LLM #LLMs #GenerativeAI #GenAI @lexfridman @KirkDBorne @Ronald_vanLoon @erikbryn @antgrasso @sallyeaves @Nicochan33 @HaroldSinnott @mvollmer1 -
Verification badges and community notes fail to solve misinformation problem
By
–
Verified handle doesn’t mean much anymore since anyone can pay their way to verification. Community notes is a good initiative but only a partial solution to the fake news problem.
-
22 Countries Endorse AI Governance Framework at Summit
By
–
The Guidance Note on AI Governance, endorsed by 22 countries, was released at the #IndiaAIImpactSummit2026, providing a shared framework for nations navigating the complex regulatory and ethical dimensions of Artificial Intelligence.
— IndiaAI (@OfficialINDIAai) 11 avril 2026
Know more: https://t.co/AxxDh8ELPE… pic.twitter.com/QfvYBvrUIOThe Guidance Note on AI Governance, endorsed by 22 countries, was released at the #IndiaAIImpactSummit2026, providing a shared framework for nations navigating the complex regulatory and ethical dimensions of Artificial Intelligence. Know more: http://
impact.indiaai.gov.in/outcome-resour
ces
…