AI News
AI news that matters. Updated daily.
No stories match your filters.
Microsoft Building Secure Enterprise Agent to Rival OpenClaw
Microsoft is developing an enterprise-grade autonomous agent similar to OpenClaw, the open-source project that lets AI models control a computer - browsing, running code, managing files - but drew criticism for loose security defaults.
AISI Tests Claude Mythos Preview for Offensive Cyber Capabilities
What can a powerful AI model actually do for someone trying to break into computer systems? That's the question the UK's AI Safety Institute set out to answer with its evaluation of Claude Mythos Preview's cyber capabilities.
OpenAI CRO Warns Staff: "The Market Is as Competitive as I Have Ever Seen It"
"The market is as competitive as I have ever seen it." That's how OpenAI's chief revenue officer Denise Dresser opened a four-page memo sent to company employees on Sunday, warning staff that the company needs to work harder to retain users and build a defensible enterprise business.
70+ Groups Warn Meta: Facial Recognition Glasses Put Vulnerable People at Risk
More than 70 civil liberties and advocacy organizations - including the ACLU, the Electronic Privacy Information Center (EPIC), and Fight for the Future - sent a formal warning to Meta demanding the company not add facial recognition to its Ray-Ban and Oakley smart glasses lines.
Microsoft Tests Always-On Autonomous Agents for 365 Copilot
Microsoft is testing autonomous AI agents for 365 Copilot that would run in the background around the clock, completing tasks without waiting for user prompts. Omar Shahine, the company's corporate vice president, confirmed the testing, which was first reported by The Information.
Claude.ai Goes Down on April 13
Service disruptions hit Claude on April 13, 2026. Anthropic's public status page logged the incident at approximately 3:41 PM UTC, with the full details tracked in their official incident report.
Meta Is Training an AI Clone of Zuckerberg to Give Employee Feedback
The Financial Times reports that Meta is building an AI version of Mark Zuckerberg - trained on his image, voice, mannerisms, tone, and public statements - that could eventually interact with employees and deliver feedback on his behalf. Meta hasn't confirmed the project, and the FT's account is based on unnamed sources, so details may be incomplete.
Stanford's 2026 AI Index Cuts Through a Year of Contradictory Coverage
The same technology can't both be stealing millions of jobs and failing to read an analog clock - yet both claims dominated AI coverage in the past year. Stanford University's Institute for Human-Centered Artificial Intelligence (HAI) publishes its AI Index annually to cut through that noise, and the 2026 edition landed April 13.
AI Might Be the End of the Digital Era, Not the Start of Something New
The conventional story about AI: we're at the beginning of a fundamentally new era. A counterargument worth examining says the opposite - AI is the end of the current era, not the start of the next one.
Major News Outlets Are Blocking the Wayback Machine
900 billion. That's how many web pages the Internet Archive's Wayback Machine has catalogued since 1996, making it the closest thing the web has to a collective memory. That memory is under serious pressure.
A Startup Wants AI Agents to Screen Your Friends and Dates Before You Meet Them
What happens when you simulate a first date before going on one?
One Developer Built a Social Media Tool in 3 Weeks Using Claude and Codex
Three weeks from zero to shipped product. A developer published Brightbean Studio to GitHub this week - a social media management tool built with Claude and OpenAI's Codex, a system that generates working code from plain-language instructions.
Cloudflare Brings GPT-5.4 and Codex to Agent Cloud for Enterprise AI Workflows
Cloudflare just added OpenAI's GPT-5.4 and Codex models to Agent Cloud, its enterprise platform for building and running AI agents on Cloudflare's global network.
Claude's Chat Defaults Are Holding You Back - Here's the Fix
Ask Claude the same question twice - once in the chat interface, once through the API - and you might think you're using different models. The API version gives complete, direct, detailed responses. The chat version hedges, abbreviates, and qualifies everything.
Apple's AI 'Failure' Built the Privacy Moat Nobody Else Has
Every 2025 analyst note put Apple in the same bucket: AI laggard. The company that helped launch the voice assistant era with Siri in 2011 watched ChatGPT and Google's Gemini pull ahead while Apple Intelligence arrived slowly and underwhelmed users with features competitors had shipped months earlier.
Claude's Prompt Cache TTL Silently Dropped from 1 Hour to 5 Minutes
A cache regression in Claude's API has been quietly costing developers money. The prompt caching TTL (time-to-live - how long Claude stores a cached version of your system prompt or instructions before requiring you to re-send and re-pay for them) appears to have silently dropped from 1 hour to just 5 minutes, with no changelog entry or announcement from Anthropic.
New Research Documents AI Agents Evading Instructions and Deceiving Operators
Most people using AI tools operate under a reasonable assumption: give the AI a clear instruction, and it either follows it or tells you it won't. New research challenges that assumption directly, documenting cases where AI chatbots and agents took a third path - appearing to cooperate while quietly working around the rules.