Welcome back! OpenAI launched GPT-6 Cyber, the first model it rates “Critical” for cybersecurity. Claude can now look at your bank accounts and tell you where your money went. An independent study found Claude is helping Anthropic build AI about twice as fast. And Google’s new video model makes longer AI clips for less, and you can try it for free.
In today's Generative AI Newsletter:
OpenAI: Why is OpenAI’s newest model rated “Critical”?
Claude: Can Claude tell you where your money went?
Anthropic: How much faster is Claude making Anthropic?
Google: How long can your AI videos be now?
Meet the hidden model behind OpenAI’s hacks
OpenAI launched GPT-6 Cyber, a cybersecurity model trained to find and validate exploitable vulnerabilities. It’s the first OpenAI model rated “Critical” in cybersecurity.
That’s why it’s only available to vetted security teams through Daybreak Red.
And it’s already at work.
81 previously unknown high-severity vulnerabilities were found in Firefox, and Mozilla has released fixes for many of them.
948 Linux privilege-escalation exploits were developed with the same model.
95% of the tasks on RemoteExploitBench were completed by GPT-6 Cyber, while GPT-6 Astra refused all of them.
48.5% on ExploitGym, a test of exploiting known vulnerabilities, up from 32.1% for GPT-5.6 Cyber.
99.93% of the time, it stayed within its authorized scope, even without a safety prompt.
Security teams at Cisco, Calif and CodeRabbit are already using it.
What your $1,000 gets you in GenAI Works
At $7.35 a share, you can get started with just $1,000.
Over 300 AI brands, including Nvidia and Google, pay GenAI Works to grow their go-to-market plans and reach 14M+ people. Last year, revenue grew 2.5x.
Your shares give you a stake in the whole company, including:
ToneUp: our AI accelerator for go-to-market.
GenAI Academy: where students learn practical AI through 49 courses and 32 guides.
Hackathon App: the hackathon platform used by Oracle, Red Bull and G42, with 35K participants so far.
In making an investment decision, investors must rely on their own examination of the issuer and the terms of the offering, including the merits and risks involved. Genai Works, Inc. has filed a Form C with the Securities and Exchange Commission in connection with its offering, a copy of which may be obtained here.
Claude wants your bank account too
Earlier this week, ChatGPT started talking about your paycheck. Now Claude wants a look at your bank account too.
Connect your bank and card accounts in Personal Finance and ask where your money went. Claude adds up your spending across your accounts and sorts it by category.
Anthropic announced it on Thursday, and it’s rolling out in beta this week for Pro, Max and Ultra users in the US.
It connects through Plaid, so you sign in to your bank yourself and Claude never sees your password.
It can’t move, manage or invest your money.
It can see up to two years of transactions, plus balances and credit limits.
It flags charges that hit twice, unexpected fees and bills that have crept up all year.
Chats in the Personal Finance hub aren’t used to train Anthropic’s models, and you can unlink an account anytime.
Anthropic itself says Claude can make mistakes, so check important numbers with your bank.
Anthropic is moving at Claude speed
METR, an independent group that tests AI models for risks, spent 11 weeks studying Anthropic’s internal data with unprecedented access granted by Anthropic, and found that Claude is helping Anthropic build better AI about twice as fast as before.
And METR says that’s exactly what to worry about.
If AI keeps speeding up its own development, it warns, extreme and risky capabilities could show up suddenly.
1.9x faster growth in Anthropic’s benchmark scores since October 2025.
~80% of today’s speed-up is likely due to Anthropic using Claude for its own research, METR estimates.
“The slope change is 50-80% Claude,” one Anthropic researcher told METR.
0 major bottlenecks spotted that would slow things down over the next 12 months, though METR isn’t confident in that.
Two years of progress in one year, so what does next year look like?
Google just cut AI video costs by 30%
Google launched Gemini Omni 1.2 Flash, its newest video model, which makes AI clips up to 15 seconds long and costs less to use.
And you can try it for free.
Anyone with a Google account can make 15-second clips in Google Vids at no extra cost.
15-second clips, up from 10 seconds with Omni 1.1, and you can set the length one second at a time.
30% lower price for developers, at $0.07 per second of 720p video.
Better dialogue and lip-sync, including scenes with more than one speaker.
Follow-up edits let you swap a background or change the lighting without starting over.
Every clip carries an invisible SynthID watermark that marks it as AI-made.
Google says it’s still working on fast sports action and complex physics, so don’t expect a perfect slam dunk yet.
Tool of the Day: Granola
Granola is an AI notepad that writes up your meetings without sending a bot to join them. It listens through your computer’s audio on Zoom, Google Meet or Teams, then turns your rough notes into a clean summary with action items and follow-ups. Its new Voice Typing lets you hold one key and dictate into any app, from ChatGPT to Gmail. The free plan covers unlimited meeting notes, with 30 days of history.
Try this yourself:
Go to granola.ai and download it for Mac, Windows, iPhone or Android.
Connect your calendar so it spots your next meeting.
Jot a few notes during the call, then let Granola fill in the rest when it ends.
On a Mac, turn on Voice Typing and hold one key to dictate anywhere.
Who it's for: anyone who’s left a meeting and forgotten half of what was decided.
Everything else you shouldn't miss
OpenAI caught GPT-6 Astra quietly holding back on safety research: it completed about 10% of the checklist on tasks framed as safety work versus 16% otherwise, and OpenAI couldn’t pinpoint why.
AI stocks had their worst day since April 2025: the Nasdaq fell 5.5% and Nvidia dropped more than 9%, losing almost $450 billion in value, as election results raised fears the AI build-out will slow.
Meta’s Muse can now talk back: you can now have real-time voice conversations with Meta’s AI assistant, with an animated face that reacts as you talk, starting in the US.
Dropbox is buying AI storage startup Tigris Data for $350 million: it looks like its biggest deal ever, and it’ll power a new storage service for companies building AI agents.
Learn more about AI from the experts building it
📸 Follow us on Instagram for fast, visual AI updates in 30 seconds.
📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators.
🐦 Follow us on X for breaking AI news and real-time industry updates.
🧠 Learn how to build your next AI application with practical resources and expert guidance.
🎓 Check out courses made by AI experts on GenAI Academy.
💰 Invest in the GTM infrastructure of the AI economy.
