The Archive
Search the full wire by company, model, lab, or keyword. Every story we have ever aggregated.
SafetyKit scales risk agents with OpenAI’s most capable models
SafetyKit product leverages GPT-5 for content moderation and compliance enforcement with improved accuracy over legacy systems.
A People-First AI Fund: $50M to support nonprofits
OpenAI opens $50M People-First AI Fund applications for US nonprofits focusing on education, community innovation, and economic opportunity.
Anthropic is endorsing SB 53
Anthropic endorses California SB 53 frontier AI regulatory framework for safety and security requirements.
Why language models hallucinate
OpenAI research identifies mechanistic causes of language model hallucinations and proposes improved evaluation methods for reliability.
GPT-5 bio bug bounty call
OpenAI announces GPT-5 bio bug bounty program with up to $25K rewards for identifying safety vulnerabilities via universal jailbreaks.
OpenAI and Greek Government launch ‘OpenAI for Greece’
OpenAI partners with Greek Government to deploy ChatGPT Edu in secondary schools and support local AI ecosystem development.
Updating restrictions of sales to unsupported regions
Anthropic restricts Claude access in unsupported regions to align with democratic interests.
Anthropic Signs White House Pledge to America's Youth: Investing in AI Education
Anthropic commits to AI education through White House pledge to America's Youth initiative.
Expanding economic opportunity with AI
OpenAI launches Jobs Platform and AI Certifications to increase worker access to training and employment opportunities.
Anthropic raises $13B Series F at $183B post-money valuation
Anthropic raises $13B Series F at $183B valuation; revenue grew 5x to $5B in eight months.
Make Memory work for you.
Mistral AI adds memory and control features to AI systems, enabling transparent user recall management.
Vijaye Raji to become CTO of Applications with acquisition of Statsig
Vijaye Raji appointed CTO of Applications at OpenAI following Statsig acquisition; reports to Fidji Simo.
Le Chat. Custom MCP connectors. Memories.
Le Chat integrates 20+ enterprise platforms via MCP connectors and adds persistent memory capabilities.
Building more helpful ChatGPT experiences for everyone
ChatGPT adds parental controls for teens, routes sensitive conversations to reasoning models, and strengthens safety protections.
What exactly does word2vec learn?
Berkeley BAIR provides closed-form theory of word2vec learning dynamics, reducing representation learning to least-squares matrix factorization.
Introducing gpt-realtime and Realtime API updates
OpenAI releases gpt-realtime speech-to-speech model with Realtime API updates including MCP support, image input, and SIP calling.
Supporting nonprofit and community innovation
OpenAI announces $50M People-First AI Fund for U.S. nonprofits in education, healthcare, and research; applications Sept–Oct 2025.
Grok Code Fast 1
xAI releases grok-code-fast-1, a lightweight agentic coding model for cost-efficient code generation.
Detecting and countering misuse of AI: August 2025
Anthropic publishes August 2025 report on detecting and countering AI misuse.
Introducing the Anthropic National Security and Public Sector Advisory Council
Anthropic forms bipartisan National Security and Public Sector Advisory Council for U.S. government engagement.
Collective alignment: public input on our Model Spec
OpenAI's survey of 1,000+ people worldwide reveals alignment between public values and Model Spec; informs AI behavior defaults.
OpenAI and Anthropic share findings from a joint safety evaluation
OpenAI and Anthropic publish joint safety evaluation results on misalignment, hallucinations, jailbreaking, and instruction following.