All sections
News
507 stories · page 2 of 57
SecurityOpenAI and NATO: Daybreak enters the Alliance's cyber defence, and NATO has used up to 300 billion tokens a monthOpenAI is giving NATO's core institutions access to its cyber models through Daybreak; according to OpenAI, the Alliance has used up to 300 billion input and output tokens a month. ENISA found four vulnerabilities, CERT Polska six in MikroTik RouterOS.FrontierMicrosoft and Salesforce launch Copilotforce: Salesforce context comes into Copilot, Teams and FabricMicrosoft and Salesforce announced Copilotforce on 17 November: Salesforce context and actions come into Copilot, Teams and Fabric, and Accenture is both the implementation partner and a pilot customer.BuildersDevin Bots: about 20 bots took a game studio from 50 to 1,900 pull requests a weekCognition announced Devin Bots on 17 November: a persistent agent that holds a goal and leads a team of Devins. According to Devin, the Android Bot built an app over 44 days, and a game studio with one human and about 20 bots reached 1,900 pull requests a week.SecurityQuad9 pulls its resolver out of France, and Belgium wants Cloudflare, Google and Cisco to block by January 2027According to Cloudflare, Quad9 is withdrawing from France after a court ruling with daily penalties, and a Belgian court is threatening Cloudflare, Google and Cisco with substantial penalties if they do not block through their public resolvers by early January 2027.FrontierInstagram launches Direct: a messaging-only app, first in Türkiye and ThailandInstagram launched Direct on 16 November: a standalone messaging app being tested in Türkiye and Thailand, working with or without an Instagram account. The built-in protections come from day one, and the age limit is 13.InfraHugging Face switches its own router for OpenRouter's for open-weight modelsHugging Face announced on 16 November that Inference Providers moves from its own router to OpenRouter's. A failed request is now retried on another provider, and your own provider API key no longer works with the language models.SecurityAnthropic will cut its vulnerability disclosure window from 90 to 30 days, effective 1 January 2027Anthropic will cut its coordinated disclosure window from 90 days to 30 for high- and critical-severity issues and to 45 for the rest, effective 1 January 2027. Findings without human review, such as those from OSS Scanner, stay outside the new policy.SecurityGoogle: autonomous frameworks are already building attacks while you sleepAccording to Google Threat Intelligence Group, bad actors now use autonomous LLM frameworks that scan software updates, figure out which vulnerability was patched and build an attack for it, with no human involvement. Google is also announcing Google AI Threat Defense, which the company says combines Gemini and other frontier models, Wiz's context and prioritization, CodeMender's automated patching and Mandiant's expertise.
507 stories