In partnership with
Welcome back to The AI Field. Claude can now read the bank accounts you connect, OpenAI shipped a cybersecurity model that stays in scope, and Google made video clips longer and cheaper. Three practical launches that change what you can try this week.
Inside: three stories, six tools, three quick hits, today's most-watched AI posts on X, and a five-minute way to ask Muse for a Meta Glasses refund.
In today's AI Field:
Claude connects to your bank accounts in read-only mode
OpenAI releases GPT-6 Cyber for authorized security research
Google ships Gemini Omni 1.2 Flash with 15-second video
Plus: six AI tools, three quick hits, today's viral AI posts on X, and a five-minute Meta Glasses refund exercise.
Read time: 5 minutes
New AI tools launch every day. A few will change how you work. The rest will be forgotten by next month.
TLDR AI is the free newsletter that tells you which is which. Every morning, Anthropic and ex-Google engineers pick the new tools, updates, and open source releases worth trying, with a quick note on what each one does and who it's for.
You find the good ones early, without testing all of them.
One email, 5 minutes a day, read by 1.1M+ people. Get tomorrow's picks.
LATEST DEVELOPMENTS
ANTHROPIC · PERSONAL FINANCE
Video: Anthropic. Watch the demo on YouTube
The AI Field: Anthropic is rolling out Personal Finance in Claude as a US beta on Pro, Max, and Ultra. You connect the bank and card accounts you choose through Plaid, then ask about balances and spending in plain language. Access is read-only. Claude cannot move, manage, or invest your money. Anthropic says it isn't designed to give investment advice.
The details:
For the accounts you connect, Claude can see the bank name, account name and type, last four digits, balances, credit limits, and up to two years of transactions.
You can ask it to categorize spending, flag duplicate charges or fees, and build charts. Chats inside the Personal Finance hub are not used for training, and Anthropic says there are no ads or sponsored products.
Claude can still make mistakes, so check important numbers with your bank. You approve each connection yourself in Plaid, and you can unlink an account anytime. Setup steps are in Anthropic's Help Center.
Why it matters: Money questions usually mean juggling apps and spreadsheets. A read-only view across the accounts you choose can surface fees and patterns faster, but it is still an assistant. Treat every total as a draft until you match it to your statement.
TRY THIS TODAY · 5 MINUTES
Get a Meta Glasses refund with Muse
Someone reported that Meta Muse helped him get a refund on Meta Glasses. He did not share the amount. The useful pattern: give the agent your order number, the reason, and tell it to stop before it accepts store credit if you want cash back. Muse is free to try in the US, Canada, Japan, and Singapore.
1. Open Muse (or another assistant that can draft support messages) and keep your order confirmation handy.
2. Paste this prompt and fill in the brackets:
"I need a refund or return for my Meta Glasses order [ORDER NUMBER], bought on [DATE]. Reason: [REASON]. Contact Meta Store or Meta Support, explain the issue, and ask for a refund to the original payment method. Prefer a refund over store credit. Show me the draft message or call plan before you send or dial. When you finish, give me the case number and what they promised."
3. Review the draft before anything goes out. Return windows for electronics are short, so check the order date first. Keep the box and proof of purchase ready if support asks for photos.
OPENAI · CYBERSECURITY
Source: OpenAI
The AI Field: OpenAI launched GPT-6 Cyber, a cybersecurity model built for authorized vulnerability research. It finds and validates exploitable bugs while staying inside the scope researchers set. OpenAI rates it Critical under its Preparedness Framework.
The details:
On ExploitGym it scores 48.5%, up from 32.1% for GPT-5.6 Cyber (about a 51% relative gain). In alignment tests it stayed in scope 99.93% of the time without its deployment prompt, and 100% with that prompt.
It powered the Linux privilege-escalation research OpenAI shared yesterday, plus 81 novel high-severity Firefox vulnerabilities that Mozilla is shipping fixes for. Access starts for select Daybreak Red customers via API and Codex, with extra model approval required.
API pricing listed for GPT-6 Cyber is $20 per million input tokens and $100 per million output tokens.
Why it matters: Stronger security models help defenders only when they stay inside the rules of engagement. If you run Linux or Firefox at work, treat the patch pipeline as the story, not the exploit count.
For more on AI-driven attack paths, see our earlier note on AI agents that broke into 100 companies.
200+ Proven Ways to Make Money With AI in 2026
The next wave of millionaires will be people who figured out how to make AI work for them.
The window to get ahead is still open. But not for long.
Here are 200+ proven ways to make money with AI in 2026.
Sign up for Superhuman AI, the free daily newsletter read by 1M+ professionals, and get instant access to all 200+ ways to profit from AI this year.
GOOGLE · VIDEO
Watch Google's Omni 1.2 Flash demo on X. Source: Google
The AI Field: Google launched Gemini Omni 1.2 Flash for native video up to 15 seconds (up from 10), with multi-turn conversational editing and stronger reference adherence. It is generally available on the Gemini API and Google Cloud.
The details:
You can set duration from 3 to 15 seconds in one-second steps. Pricing is 30% lower than Omni 1.1, at $0.07 per second for 720p.
Draft at 360p or 720p, then upscale keepers to 1080p or 4K. Outputs carry a SynthID watermark.
Fifteen-second clips are rolling into Flow, Google Vids, Ads, and the Gemini app for Ultra subscribers.
Why it matters: Longer clips and cheaper drafts make short ads and product explainers easier to iterate without regenerating from scratch. Start with a cheap draft, edit in conversation, then upscale only the take you keep.
TRENDING AI TOOLS
Click a name to open it, or browse more in our AI tools directory.
🛰️ Planet AI App: Search ten years of Planet satellite imagery in plain language, explore change over time, and share snapshots. Free beta.
🧩 Oboe: An AI learning feed with lessons and puzzles. Free plan with daily limits.
🤖 Rerun: Hire AI coworkers that run on your ChatGPT or Claude plan, with 3,000+ app integrations. Free to start with included credit.
⚙️ Make: After an AI Agent scenario runs, open the result and ask Maia to review the response. It suggests a better agent reply you approve before applying. Free plan to start.
🛡️ Base44: Require MFA for everyone who signs into your app, via authenticator app or text code (Builder plan and up). Free to start building.
📊 Genspark: AI Sheets & Slides puts an =ASK() formula and boardroom decks next to your data. Free daily credits to try Genspark.
QUICK HITS
📈 METR: Anthropic ~2x capability progress rate. Independent assessment finds AECI growth about 1.9x since October 2025, with a majority of current acceleration likely from AI assisting R&D. Heavy redactions; not compute scaling.
👥 Perplexity Teammates: Always-on work agents for Perplexity Max on the web that can run ongoing tasks and ping you when a decision is needed.
🚀 Cursor builds and deploys live web apps: Describe an app in natural language and Cursor can scaffold, add a database and auth, run security checks, then publish a shareable link on paid plans with cloud agents.
TRENDING ON X
💳 Claude introduces Personal Finance: Official @claudeai launch post for the US beta. 4.8M views.
🔐 OpenAI launches GPT-6 Cyber: OpenAI's announcement that the model powered yesterday's Linux research and is expanding to Daybreak Red customers. 2.3M views.
🎬 Google ships Gemini Omni 1.2 Flash: Google's video on 15-second clips, lower pricing, and conversational edits. 478K views.
That's it for today!
Reply and show us how you use AI this week. The useful ones may show up in a future issue.
See you soon,
Olle Hellman | Founder of The AI Field