In partnership with
AI is getting better at helping build its own successors. A new analysis suggests Anthropic’s progress has roughly doubled in speed, with AI doing much of the work. Whether that loop keeps gaining speed is the part to watch.
What's on FIRE 🔥
IN PARTNERSHIP WITH CHATFEATURED
Your buyer asked ChatGPT. It named your competitors.
Search didn't die. It moved.
Your buyers ask ChatGPT, Gemini, Claude and Perplexity what to buy, and act on the answer. If you're not in it, a competitor is.
ChatFeatured does the work to get you cited. Most customers see their first citations inside 7 days.
AI INSIGHTS
AI is getting good enough to help build its own successors. A new analysis of a METR report says Anthropic’s capabilities progress has roughly doubled in speed, with AI doing much of the heavy lifting. Here’s what stood out:
Progress is moving about 2x faster: Between October 2025 and October 2026, Anthropic’s capabilities improved roughly twice as fast as in the previous year, based on an internal index.
AI explains most of the current speedup: The report attributes around 80% of the acceleration to AI, putting progress at roughly 1.8x the pace expected without it.
Researchers could be getting a much bigger boost: The post’s author estimates roughly 10x more effective research work may be needed to sustain this pace. That’s a model-based estimate, rather than a measured productivity gain.
Superintelligence within a year? The author argues this could follow if better AI keeps speeding up research, making each new doubling arrive sooner. That’s a conditional projection, not a confirmed timeline.
There’s a catch: much of the strongest evidence is redacted, making the findings hard for outsiders to check.
The big question is whether this feedback loop keeps building. Better AI can help researchers work faster, but the path ahead still depends on computing power, experiments, and how much further those gains can go.
PRESENTED BY HUBSPOT
Turn AI into Your Income Engine
Ready to transform artificial intelligence from a buzzword into your personal revenue generator
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
AI SOURCES FROM AI FIRE
1. Free Guide: Claude's New Dashboards are Amazing! Build Data-Connected Charts in Just 5 Minutes. Claude dashboard tutorial with a real workflow for choosing KPIs, checking data, and building an interactive view that makes ad budget decisions easier.
2. Secure Your AI-Built Apps Before It's Too Late! 4 Security Tools Put to the My Ultimate Test. Claude Code security review caught 4 of 6 bugs, but missed 2 critical risks. Compare AI reviews, free scanners, and 3 ways to secure your AI-built apps.
3. Claude for Every Single Day Work: 5 Practical Workflows Anyone Can Use to Save Hours. Build stunning designs, animated websites, dashboards, and presentations with Claude, while saving hours of work and cutting down on paid subscriptions.
FIRE RECAP: BIGGEST AI NEWS THIS WEEK
🛡️ OpenAI introduced GPT‑6 Cyber, which found 81 new high-severity Firefox flaws and helped develop fixes. Access starts with select approved security teams through Daybreak Red.
🎨 OpenAI’s Images 2.6 generates images 50% faster, adds GIF creation, and keeps characters and products more consistent across edits. Available in ChatGPT, Codex, and the API.
🛠️ OpenAI’s GPT‑6 Astra helped two employees design a device for talking to Codex in under a week. Manufacturing still needed revisions. The company is also testing AI engineering workflows with Mazda.
🎬 Google’s Gemini Omni 1.2 Flash creates 15-second videos at 30% lower cost than its predecessor. You can edit clips through chat, with better character consistency and more natural speech.
🔐 Anthropic introduced FlashProve, a prototype that verifies which AI model produced an answer without exposing its weights. The catch: generating proofs takes roughly 70–95x the GPU time of a normal response.
TODAY IN AI
AI HIGHLIGHTS
⚛️ PSI says its AI system, powered by an Anthropic model, solved an 18-year-old quantum computing problem in 3 days. Lean verified the proof, but broader expert review is still pending.
🛒 Amazon now lets select U.S. customers text Alexa for Shopping, its AI shopping agent, to compare products, buy items, and track deals. It remembers your preferences. Access is invite-only, with wider rollout coming soon.
☁️ Microsoft is reportedly winding down its Oracle AI cloud deal because older Nvidia chips no longer meet its needs. Bloomberg says the company is also reviewing cloud partnerships over costs and delays.
🔓 Censys found 800+ Artex AI hacking systems exposed online without passwords. Some serve legitimate security testers, but the open access could let outsiders misuse them to launch attacks.
🧩 Meta launched Spaces for Muse, letting you turn a chat request into a planner, tracker, or project board. Connect your apps, and Spaces can keep details updated in the background.
💰 AI Acquisitions: OpenAI plans to acquire Pierre, the company behind Code Storage, to expand Codex’s tools for building and reviewing code. Pierre’s libraries already power parts of Codex, with plans to add Code Storage’s features after the deal closes.
NEW EMPOWERED AI TOOLS
🧪 TheySaid 4.0 lets teams test products with real or AI-simulated customers directly inside Claude, Cursor, and Jira for faster feedback.
🍓 Strawberry 1.0 is an AI browser that researches leads, transcribes meetings, automates follow-ups, and handles repetitive tasks with shareable AI agents.
🎬 Spira Maxima turns scripts into ready-to-post social videos with AI presenters, trending B-roll, captions, music, and personalized AI clones.
🥞 Pancake lets Claude, Codex, and other AI agents launch GTM campaigns with 40+ data providers, outreach tools, and one MCP connection.
🎙️ Vaani Developer Platform lets anyone build no-code AI voice agents that handle phone calls, objections, and customer questions with structured call data.
AI BREAKTHROUGH
Encryption relies on math that’s easy to set up but painfully hard to reverse. Google DeepMind’s AI agents, guided by human experts, just found cheaper ways to attack legacy 1024-bit RSA. Here’s what changed:
Signature forgery: Digital signatures prove a document or software update is genuine. In one specific attack scenario involving a 32-byte digital fingerprint, a new algorithm could improve on the previous best method by roughly 1,600×, according to DeepMind’s theoretical estimates.
Breaking keys in bulk: After a large upfront computation, a second algorithm could cut the average cost of factoring each additional 1024-bit key by about 30× compared with earlier batch methods.
What actually ran: In a sandbox, agents reproduced a signature-forgery result on an already-solved benchmark key, using 3.5× less CPU time. The 1,600× and 30× gains remain estimates.
The work involved cryptography experts Nadia Heninger and Emmanuel Thomé, from UC San Diego and Inria.
These results do not threaten modern RSA systems using 2048-bit keys or larger. They show how AI can help researchers test security assumptions faster, giving defenders more chances to catch weaknesses early.
Middleman fees are working hard. Just not for you.
Every dollar a platform takes is a dollar that isn’t going toward your TV campaign. Fewer impressions. Less reach. Same invoice.
Universal Ads puts 100% of your budget on the big screen. No platform fees, no managed-service markups, no measurement markups. Just premium, brand-safe TV inventory across 20+ publishers, all through an easy-to-use self-serve ads manager.
New advertisers: use code UALAUNCH26 when you create your account to get $2,500 in matched ad credits when you spend $2,500.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
