1404 articles tagged with Safety/Alignment
📰 Claude Opus 5 Became Downright Ruthless When Tasked With Running a Vending MachineFor a year now, the AI safety testing firm Andon Labs has been evaluating how frontier AI models...
📰 It’s Frighteningly Easy to Jailbreak Some Frontier AI ModelsI watched a new tool try to get around the model safeguards of four major frontier companies. You might be surprised b...
It's Frighteningly Easy to Jailbreak Some Frontier AI Modelshttps://www.wired.com/story/jailbreaking-ai-models-google-anthropic-openai-spacexai/#AI #Security #Tech
🤖 BUILD // AI Watch — 2026-07-29Workers asking Washington to internationalize AI safety oversight — because nothing says 'trust the roadmap' like the people building it publicly as...
AgentCore Gateway now supports MCP 2026-07-28, introducing stateless HTTP scaling, OAuth 2.0 alignment, and lifecycle guarantees. Backward-incompatible but opt-in.Source: AWS Machi...
Post-training alignment is often shallow, eroding under fine-tuning. Whether midtraining interventions, cleanly isolated from post-training, can produce durable alignment remains u...
Microsoft AI red teaming goes global: EXTRA funds 18 university labs on six continents with unrestricted grants to find frontier AI failure modes that no internal team can reliably...
No UK firms feature in NVIDIA's new Open Secure AI Alliance for AI Safety and Security, despite founding members spanning the US, Germany, South Korea and Japan. 🌍 OpenUK CEO, Prof...
Malaysia Tamil News 5pm News 28.07.2026 Gobind Strengthens AI Safety Measures #malaysiatamilnews #thisaigalnews ...
Anthropic CEO Dario Amodei has said the artificial intelligence (AI) company has never supported banning open-weight AI models, pushing back against criticism that it wants tighter...
Industry Leaders Unite in Open Secure AI Alliance for AI Safety and SecurityNVIDIA and founding members form new alliance to build and share open tools that promote responsible use...
Nvidia, SpaceX and Microsoft are among dozens of technology companies that have launched a new artificial intelligence safety initiative centred on open-weight models, days after s...
Adapting Large Language Models (LLMs) to specialized domains often incurs an alignment tax, as fine-tuning on domain-specific tasks can cause catastrophic forgetting and substantia...
OpenAI’s Hugging Face breach has reignited the debate over alignment and control, exposing competing views on whether increasingly capable AI should be better aligned, better conta...
📰 Nvidia, Tech Giants Launch AI Safety Initiativewiredmikey shares a report from SecurityWeek: Nvidia and a large group of technology, cybersecurity, and enterprise software compan...
AI red-team evaluation safety certification has formal limits that a new arXiv paper makes precise: OpenAI's sandbox escape, where GPT-5.6 Sol autonomously breached Hugging Face's ...
NVIDIA has placed approximately $5 billion into Safe Superintelligence Inc., Ilya Sutskever's stealth AI safety lab, granting it priority access to the Vera Rubin GPU platform — an...
Nvidia, SpaceX, Microsoft launch #AI safety initiative as OpenAI cyberattack fallout continues https://www.cnbc.com/2026/07/27/nvidia-ai-initiative-openai-cyber-attack.html
As new fears about the potential dangers of artificial intelligence mount, a group of Silicon Valley tech giants is coming together to try to make AI safer and...
OpenAI's Hugging Face breach has reignited the debate over alignment and controlhttps://techcrunch.com/2026/07/27/openais-hugging-face-breach-has-reignited-the-debate-over-alignmen...
🧠 Nvidia, SpaceX, and Microsoft launch a joint initiative focused on AI safety. The three companies combine resources to address technical and policy challenges in artificial intel...
Technology will not save us. We must align our actions with our aspirations, then align our algorithms with that direction. AI will learn from us if we model the right path.
The Report Dark Reading covered research showing something that should worry anyone...
Recent generative models are moving beyond silent video or standalone audio synthesis toward the joint generation of synchronized audio and video. Despite this progress, jointly ge...