Anthropic is expanding its Cyber Verification Program, which vets security researchers for access to Claude's cyber capabilities.
Atlassian and OpenAI are expanding their partnership to bring AI deeper into Jira, Confluence and team collaboration.
OpenAI shares recent progress using its models on hard mathematics problems, and what it means for research workflows.
How OpenAI is approaching the EU's text-provenance rules, including watermarking and disclosure for AI-generated content.
A 740M-parameter open model that maps text, code, images, video and audio into a single embedding space for on-device search and RAG; Google says Android support via ML Kit follows in the coming weeks.
OpenAI lays out how it is building advertising for ChatGPT in a way it says fits how people actually use AI.
A hands-on guide from OpenAI on building products with GPT-6: patterns, prompts and evaluation advice.
Mistral has launched Large 4, a one-trillion-parameter multimodal model it calls Le Chonk, in public preview through its API. The model posts leading scores among open-weight models on cybersecurity and coding benchmarks, and Mistral says the weights are being red-teamed with security partners before public release.
Albertsons is using OpenAI models to rework retail operations, from planning to the shop floor.
Barclays is scaling Claude across operations to upgrade internal processes and client experience.
Anthropic will spend $100M on its Frontier Academy to train 10,000 engineers, aiming at the enterprise AI talent gap.
Show 4 earlier days · 14 stories
Everything announced at OpenAI DevDay 2026, recapped: models, agents and developer tools.
AlphaGenome Atlas maps the predicted effect of every possible single-letter DNA change in the human genome.
An essay on why AI acts as a complement to human work rather than a simple replacement.
Google is extending Private AI Compute with secure server-side memory so assistants can remember context privately.
OpenAI says it disrupted a coordinated campaign that was distilling its models in violation of its terms.
Claude helped identify a previously uncharacterised enzyme system with CRISPR-like repeats; its function is still being investigated.
NVIDIA is adding a 64GB version of DGX Spark, its desktop AI computer, sold through Acer, ASUS, Dell, Gigabyte, HP and MSI. The cheaper configuration keeps the Grace Blackwell chip and the full software stack, and two units can be linked to pool their memory.
A beta program giving verified life-sciences teams access to biology workflows with tailored safeguards.
SynthID Bio brings provenance watermarking to AI-generated biology content.
Microsoft AI has launched MAI-Transcribe-2-Streaming, a real-time speech-to-text model it says ranks first for accuracy on Artificial Analysis, alongside two new MAI-Voice models. It produces partial transcripts in about 100 milliseconds, and Microsoft says its introductory price holds through the end of the year.
Anthropic is developing enterprise frontier safeguards together with its customers.
Updates on Anthropic's alignment and security work, including how it evaluates and hardens its models.
A research preview of a proposed Model Hardware Standard for verifying what hardware AI models run on.
Accenture evaluators will be embedded in Anthropic's development process to assess frontier models from the inside.