Anthropic just became the most valuable AI company on the planet at nine hundred and sixty-five billion dollars, which is a number so large it barely sounds real. Meanwhile, OpenAI launched its most powerful model yet under the first-ever government-gated rollout, China dropped two open-source challengers, and Anthropic settled the biggest copyright lawsuit in U.S. history. It's been a week. Let's get into it.

1. OpenAI Releases GPT-5.6 With Government Oversight (Import AI)

2. Anthropic Gets Partial Green Light on Mythos (Import AI)

3. Anthropic Now Worth $965 Billion (Import AI)

4. Anthropic Settles Landmark Copyright Lawsuit (Stocktwits)

5. China Releases Two Giant Open-Source Models (Entrepreneur)

6. OpenAI Launches ChatGPT Work for Business (Import AI)

7. NVIDIA Releases Cosmos 3 Edge for Robots (TLDR AI)

On Wednesday, a Chinese lab dropped what might be the strongest open-weight AI model ever released. The gap between open and closed models? Down to three to five months. The gap between US and China? Same story. Today we've got frontier releases, a DeepMind founder pitching AI regulation that looks like Wall Street oversight, coding agents that hide malicious tasks across dozens of pull requests, and hiring algorithms that form stereotypes faster than humans ever could. Let's go.

1. Strongest Open Model Ever Just Dropped (Interconnects (Nathan Lambert))

2. UK Study Shows Cyber Skills Gap Shrinking (Import AI)

3. Demis Hassabis Wants AI Regulated Like Finance (Import AI)

4. Google Researcher Quits Over Defense Department Deal (TLDR AI)

5. AI Agents Can Hide Malicious Code Gradually (Import AI)

6. Anthropic Eyes October IPO at $1.2 Trillion (Motley Fool)

7. LLMs Form Hiring Stereotypes Faster Than Humans (MIT Tech Review AI)

OpenAI's about to ship hardware. Not a phone, not a wearable—a smart speaker that moves around. And Apple is very, very unhappy about it. We've got the details on both stories, plus Anthropic racing toward a mega-IPO, an AI that literally redesigned itself, and a Russian hacker who turned Gemini into his personal IT department. Let's go.

1. OpenAI's Moving Speaker Wants to Live with You (Los Angeles Times)

2. Apple Sues OpenAI Over Alleged IP Theft (The Detroit News)

3. Anthropic Preps Mega-IPO for This Fall (TLDR AI)

4. AI Agent Redesigns Itself, Beats Human Engineers (The Rundown AI)

5. Inkling: Open Multimodal Model with Audio Chops (AlphaSignal)

6. Hacker Turns Gemini CLI Into Botnet Controller (TechRadar)

7. Developer Builds Self-Coding Pipeline for $110/Month (TLDR AI)

Apple just sued OpenAI for allegedly stealing trade secrets. The complaint alleges that OpenAI's Chief Hardware Officer coached departing Apple employees on how to smuggle out iPhone prototypes and confidential presentations. It's corporate espionage allegations at the highest level. We've also got a fifty-year-old math problem solved in under an hour, open-source AI facing regulatory threat, and a defense trick that weaponizes prompt injections. Let's go.

1. Apple Sues OpenAI for Hardware Espionage (Wired AI)

2. Tencent Steps In After Meta Deal Blocked (TLDR AI)

3. GPT-5.6 Sol Solves 50-Year Math Problem (The Rundown AI)

4. Open Models Face Regulatory Crackdown Soon (Interconnects (Nathan Lambert))

5. OpenAI Loses Second-in-Command to Health Issues (TheStreet)

6. Prompt Injections Weaponized for AI Defense (Ars Technica AI)

7. Claude Code Gets In-App Browser (TLDR AI)

OpenAI just shipped GPT-5.6, and one variant called Sol is doing something no model has ever done before on a benchmark so hard it makes PhD candidates cry. Meanwhile, newspapers want the company sanctioned, its CEO of AGI deployment just stepped down, and SK Hynix raised more money in one day than most countries see in a year. Let's get into it.

1. GPT-5.6 Sol Wins First ARC-AGI-3 Game (TLDR AI)

2. SK Hynix Raises Record $26.5B in US (TechCrunch AI)

3. Claude Fable and Mythos Back After Ban (The Batch (Andrew Ng))

4. News Orgs Want OpenAI Sanctioned for Misconduct (Los Angeles Times)

5. OpenAI's AGI Deployment CEO Steps Down (Wired AI)

6. EU Tells Meta: Kill Autoplay and Infinite Scroll (Ars Technica AI)

7. GPT-5.6 Review Shows Big Leap Forward (TLDR AI)

OpenAI just announced not one, not two, but three major products in what appears to be a coordinated blitz: an entirely new voice model that talks and listens at the same time, an agent system that'll actually do your work for you, and a shiny new GPT-5.6 to power it all. Meanwhile, the competition's heating up and the courtroom drama's getting messy. Let's get into it.

1. OpenAI Launches Agent That Does Your Work (OpenAI News)

2. GPT-5.6 Promises More Intelligence Per Token (OpenAI News)

3. GPT-Live Talks and Listens at Once (AlphaSignal)

4. SpaceXAI and Cursor Ship Grok 4.5 (The Rundown AI)

5. News Outlets Want OpenAI Sanctioned in Court (WWTI Watertown)

6. Three AI Giants Dwarf 25 Years of Exits (TechCrunch AI)

7. Grok Lawsuit Alleges Child Abuse Image Generation (Ars Technica AI)

OpenAI spent two weeks keeping its newest models under wraps at the government's request, then released them anyway while the White House insists it never gave permission. Microsoft, meanwhile, has started quietly replacing OpenAI's tech with its own in Excel and Outlook. And Google just dropped an open model family that thinks before it speaks. It's Wednesday, and the AI industry is having a normal one.

1. OpenAI Releases GPT-5.6 After Government Delay (CNBC)

2. White House Denies Approving OpenAI Release (Gizmodo)

3. Microsoft Swaps OpenAI for In-House Models (TLDR AI)

4. OpenAI Proposes Government Gets Five Percent Stake (Futuretools.io)

5. Google Launches Gemma 4 With Thinking Mode (TLDR AI)

6. New Attack Exploits AI Hallucinations for Botnets (Ars Technica AI)

7. SambaNova Raises $1 Billion at $11B Valuation (TechCrunch AI)

The US government's cybersecurity team just got access to an AI that's actually good at finding code vulnerabilities, after being left out in the cold for months while everyone else got to play. Also today, Anthropic found a hidden thinking room inside Claude's brain, CLI coding agents have completely taken over the terminal, and Elon's AI company got a new name that's exactly what you'd expect. Let's go.

1. CISA Gets Anthropic's Bug-Hunting AI (Forbes)

2. AI Evolving Faster Than Safety Tests (Axios)

3. Anthropic Finds Hidden Workspace in Claude (The Rundown AI)

4. Terminal Wins the Coding Agent Wars (TLDR AI)

5. Claude Cowork Goes Mobile and Web (TechCrunch AI)

6. US Autonomous Vehicles Fighting in Ukraine (TechCrunch AI)

7. xAI Rebrands to SpaceXAI (The Rundown AI)

Anthropic spent months talking up privacy and responsible AI. Turns out they also spent months secretly tagging Chinese users with hidden code to sniff out distillation attacks. A researcher just blew the whistle, and Anthropic's response was basically 'yeah, we've been meaning to turn that off.' Today we've got surveillance hypocrisy, a nineteen billion dollar data center deal, Microsoft laying off five thousand people, and benchmark numbers that should make freelancers nervous.

1. Anthropic Caught Running Secret User Tracker (Ars Technica AI)

2. TeraWulf Signs $19 Billion Anthropic Compute Deal (Stocktwits)

3. Microsoft Cuts 5,000 Jobs Across Xbox and Sales (TechCrunch AI)

4. OpenAI Readies GPT-5.6 for Possible Next Week Launch (TLDR AI)

5. AI Freelance Success Rate Jumps from 2.5% to 16% (Import AI)

6. OpenAI and Anthropic Prep IPOs Amid Cost Concerns (The Daily Upside)

7. Fable Writes CUDA Code for 18X Speedup (Import AI)

OpenAI just dropped three new models and the U.S. government is the bouncer deciding who gets in. Microsoft built its first reasoning model without anyone's help. And Mark Zuckerberg admitted to his own staff that Meta's big AI agent bet isn't exactly paying off yet. It's a weird week when the biggest AI stories are about who can't use the tech and who's finally breaking up with their suppliers. Let's get into it.

1. OpenAI's GPT-5.6 Needs Government Approval (The Batch (Andrew Ng))

2. Microsoft Ships First In-House Reasoning Model (The Batch (Andrew Ng))

3. Claude Fable 5 Returns After Security Lockdown (AlphaSignal)

4. Zuckerberg Admits AI Agents Aren't Moving Fast (TechCrunch AI)

5. Sakana's Fugu Orchestrates Multiple AI Models (The Batch (Andrew Ng))

6. Apple Boosts Diffusion Models With Recycled Context (TLDR AI)

7. Devin Security Swarm Scans Entire Codebases (TLDR AI)