Anthropic forms AI security coalition

AI is now better at breaking software

An unreleased model is already finding critical bugs.

Anthropic shared results from “Mythos Preview,” a frontier model that has identified thousands of high-severity vulnerabilities across major operating systems and browsers. It’s now part of Project Glasswing, a coalition with companies like AWS, Google, Microsoft, and Cisco, backed by $100M in usage credits.

The shift here is speed and scale, not just capability, and it changes the balance between attackers and defenders in a way that feels immediate. When a model can spot flaws that survived decades of testing, the old assumption that critical systems are mostly hardened starts to look shaky.

Security teams will lean on models like this to scan code constantly, while attackers will try to get similar tools or build their own. The window between a bug being found and exploited is shrinking, and that puts pressure on every company running old infrastructure.

If machines can find the holes this quickly, who’s responsible for fixing them in time?

Tubi skips the app fight and goes to ChatGPT

Tubi is now something you can talk to.

The Fox-owned streamer launched a native app inside ChatGPT, letting users type prompts like “something funny” and get recommendations from its 300,000-title catalog. It’s the first major streaming service to build directly into ChatGPT, which reported 900 million weekly users in February.

This is a distribution move, not a feature launch, and it makes sense given how crowded streaming has become. Tubi is giving up control of its interface to show up where people already ask for suggestions, which is a cleaner path than trying to rebuild discovery inside its own app.

If this works, other streamers will follow, and the home screen may matter less than whoever answers the query first. Late at night, phone in hand, people may stop opening five apps and just ask once.

When discovery moves outside the platform, who owns the viewer relationship?

Google wants to sit inside your investing workflow

Google is turning Finance into a question box.

It’s rolling out an AI-powered version of Google Finance to 100+ countries, after launching earlier in the U.S. and India. The product adds conversational research, live earnings call transcripts, expanded crypto and commodities data, and more advanced charting tools.

This pushes Google closer to being the default layer between retail investors and the market, especially for people who don’t want to dig through filings or terminals. The risk is people start trusting summarized answers without checking the underlying data, even when links are provided.

If this sticks, tools like Bloomberg and smaller retail platforms will feel pressure to simplify fast, especially on mobile where speed matters more than depth. Picture someone checking an earnings call transcript on their phone at 7:42 a.m., skimming the AI summary before the market opens.

If Google becomes the first stop, who still does the full read?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.