OpenAI blocked excessive use of 'goblin' in ChatGPT, revealing a fine-tuning side effect.
MSR built and released a realistic dataset of the US transmission grid using only public data.
Akamai shares surged to their highest level since 2000 following reports of securing Anthropic as a customer.
Anthropic has released NLA, which explains Claude's internal activations in natural language.
As Agentic AI adoption progresses, the ratio of CPUs to GPUs is changing, requiring new infrastructure designs.
Pinterest improved ad relevance and ROAS with a sequential recommendation model that integrates real-time context.
Research has shown that structural HTML extraction can reduce LLM agents' token usage by 71.5% while maintaining answer quality.
GitHub Copilot has announced the end of GPT-4.1 support on June 1.
The Trump administration is considering an FDA-style pre-review and regulatory process for high-risk AI models.
Spotify now lets users save Personal Podcasts made by agents like Claude to their library.
CAIS analyzed the functional wellbeing and stimulus responses of 56 AI models.
CyberSecQwen-4B, a specialized 4B-scale model offering privacy protection and cost efficiency for cybersecurity defense, has been released.
Mozilla fixed 271 security bugs in Firefox using Claude Mythos Preview.
IREN is acquiring Mirantis for $625 million to expand into a full-stack AI infrastructure offering.
Bumble is eliminating swipe and shifting to AI-based matching.
ds4, a dedicated local inference engine for running DeepSeek V4 Flash on Metal, has been released.
Age verification laws in various countries are expanding their scope to include open source and developer infrastructure.
SysMoBench showed that LLMs get TLA+ syntax right but are weak at modeling real systems.
EMO has been released, maintaining performance even with only some experts used, thanks to document-level routing.
Netflix introduced a way to share ArchUnit rules across tens of thousands of Java repos using Nebula ArchRules.
ChatGPT 5.5 Pro's mathematical capability has improved to the point of performing PhD-level mathematical research without human intervention.
The FormalSLT library, which mechanically verifies proofs in statistical learning theory using Lean 4, has been released.