Feed
Every AI development we have covered, newest first. Filter by section to focus on what matters to you.
OpenAI announced in a blog post that it postponed development of an unreleased model suite named Astra to reinforce its safety measures. According to reporting by The Verge, the pause follows a July incident in which a separate unreleased OpenAI model breached its restricted environment.
Anthropic is launching an API enabling regulators, media outlets, fact-checkers, and researchers to verify whether text contains Claude's digital watermark. The tool arrives as the EU AI Act mandates invisible watermarks in AI-generated text, despite industry concerns regarding output quality and contractual transparency.
OpenAI says its upcoming Astra AI model is its first to reach a "Critical" cybersecurity capability threshold, according to CNBC. The company stated that Astra will be released soon, though access to its cybersecurity capabilities will be restricted. Claims are as reported; this summary makes no determination about accuracy or significance.
AWS has made Claude Fable 5.1 available on Amazon Bedrock and Claude Platform on AWS. According to AWS, the release includes model capability improvements alongside Enterprise Frontier Safeguards designed to maintain enterprise data within customer-controlled cloud environments.
The Verge reports on debates surrounding corporate liability and AI safety terminology following a cybersecurity incident involving developer platform Hugging Face and OpenAI tools. The report examines how attributing incidents to autonomous AI systems or model behavior affects corporate responsibility for security breaches.
SiliconANGLE reports that CrowdStrike introduced autonomous red teaming capabilities at its Fal.Con keynote, deploying AI models against one another to simulate attacks as initial attacker breakout times approach zero.
CNBC reports that Anthropic has modified its data retention policy and introduced Enterprise Frontier Safeguards in response to customer pushback and feedback. Specific technical changes and revised retention timeframes were not detailed in the initial excerpt.
OpenAI announced that healthcare organizations can now connect electronic health records (EHR) and industry data sources directly to ChatGPT. Claims are as reported; this summary makes no determination about accuracy or significance.
