Published today, sourced from stories within the last 24 hours.
🔒 AI-Found PHP Flaws Trigger Global Emergency Patching
The internet’s biggest story right now: two critical PHP vulnerabilities (CVE-2026-109618 and CVE-2026-109619) forced emergency patching worldwide, with Finland temporarily shutting government web services and UK councils and the BBC taking pages offline. The flaws allow unauthenticated remote code execution on virtually every web-serving PHP installation. Notably, AI played a central role in finding and weaponizing the bugs — OpenAI says its GPT-5.6 Sol model identified one flaw, while Wiz independently found both. OpenAI is calling the gap between AI-assisted discovery and public disclosure a “Defender’s Window,” a sign that security disclosure timelines are being compressed by frontier models.
🚀 Anthropic Opens Claude Duet Research Preview
Anthropic launched Claude Duet, a proactive agent built on Claude Opus 5.5 that reaches out before you ask. Running on web, iOS and Android, Duet surfaces work it thinks deserves attention and can continue tasks after you step away — Anthropic says it asks before acting on your behalf. The research-preview waitlist is open now for Pro and Max subscribers in the US.
☁️ OpenAI Frontier Models Coming to Google Cloud
In a landmark partnership announced this week, OpenAI and Google Cloud said OpenAI’s frontier models and its Codex agentic workflows will land in Google Cloud’s Model Garden, purchasable through the Google Cloud Marketplace with existing spend commitments applying. General availability is planned in coming weeks — putting OpenAI’s technology on sale through a direct competitor’s cloud for the first time.
💰 DeepSeek Raises ~$15B at $74B Valuation
Chinese AI lab DeepSeek has raised approximately $15 billion in a massive round backed by Tencent, CATL and other Chinese tech and industrial giants, per reports published today. The raise gives the Hangzhou-based company major new firepower to challenge US labs — while Reuters notes a new inexpensive Chinese model is catching up with Anthropic and OpenAI on their home turf.
🤝 Mistral and NVIDIA Team Up on Industrial Engineering AI
Mistral and NVIDIA announced plans to connect Mistral models with NVIDIA simulation and computing tools for engineering design, including public datasets and benchmarks testing whether AI agents can complete computer-aided engineering and chip-design work accurately.
🏛️ Anthropic Proposes Federal AI Safety Legislation
Anthropic put forward a federal safety bill proposal that would let the Commerce Department seek court-ordered suspension of AI models that cross specified catastrophic-risk thresholds. Critics question evaluator independence and intervention powers below those thresholds. Separately, Anthropic CEO Dario Amodei publicly defended the company’s “race to the top” strategy against criticism that safety has become secondary to winning the frontier race, citing a four-month delay of its Mythos model as a commercial cost the company accepted for safety reasons.
🪖 Google DeepMind Staff Resign Over Pentagon AI Work
BBC reporting today highlights resignations at Google DeepMind over military AI work, with employees questioning how much visibility and control Google has over how the Pentagon uses its AI. Google says public-sector contracts go through compliance reviews against its AI principles.
📊 Meta Doubles Down on AI Spend After Q3 Earnings
Following its Q3 report, Meta raised its 2026 capex floor to $135B and created a non-voting Class C stock to keep funding AI buildout. CEO Mark Zuckerberg said more than 10 million people have set up a Muse, with over 25% making purchases through the app at an average size above $200.
🧪 Also Today
- Recursive releases Flint: Richard Socher’s startup shipped a local coding model built by an automated AI research system — 39.8B total parameters, ~2.5B active per token, exceeding 125 tokens/sec on a MacBook Pro. Checkpoint released.
- Perplexity’s JOLT: Researchers trained an AI “teacher” model that adapts lessons to its student model, with reported gains across math, coding and agent tasks.
- OpenAI’s Orchestra Bio acquisition pushes the company deeper into AI-driven drug development; Claude-designed toxin binders moved into lab testing.
- Demis Hassabis now expects AGI matching the full cognitive range of the human brain within two to three years.
- Youth-led group Design It For Us demands Congress pass AI protections for young users by end of 2027, based on input from 700+ members across 40 states.
⚠️ Disclaimer
Published today, sourced from stories within the last 24 hours. Details are drawn from press releases and news reports and may evolve as stories develop.
Leave a Reply