Mythos Unleashed: How Mozilla Turned AI Bug Hunter into Firefox’s Biggest Security Overhaul

Mozilla harnessed Anthropic's Mythos Preview to patch 271 Firefox bugs in version 150, building on 22 prior Claude finds. AI now scans code exhaustively, forcing a security reckoning for open source amid attacker access risks.
Mythos Unleashed: How Mozilla Turned AI Bug Hunter into Firefox’s Biggest Security Overhaul
Written by Ava Callegari

Firefox just shipped its toughest security upgrade yet. Version 150 patches 271 vulnerabilities uncovered by Anthropic’s Mythos Preview, a model so potent its creators handed early access only to a select few. Mozilla’s team, led by chief technology officer Bobby Holley, calls it a bootcamp for software. Every line of code now faces automated scrutiny that mimics—and surpasses—human experts. But this flood of flaws demands discipline. Resources. Grit.

Bobby Holley puts it bluntly: “Our belief is that the tools have changed things dramatically, because now we have automated techniques that can cover, as far as we can tell, the full space of vulnerability-inducing bugs.” For years, Firefox relied on fuzzing and manual hunts. Attackers did the same, but only if they shelled out millions. No more. AI collapses that barrier.

The collaboration started earlier with Claude Opus 4.6. In two weeks, it scanned nearly 6,000 C++ files, filed 112 reports, and flagged 22 security bugs—14 high-severity, nearly a fifth of 2025’s total fixes. Mozilla rushed patches into Firefox 148, as detailed in their blog post. Anthropic confirmed the haul in their announcement, noting one use-after-free in the JavaScript engine caught after just 20 minutes (Anthropic).

That was the warmup. Mythos Preview arrived with sharper teeth. It doesn’t just spot bugs. It exploits them. On Firefox 147’s JavaScript engine—flaws already patched in 148—Opus 4.6 succeeded twice in hundreds of tries. Mythos nailed 181 working exploits, a 90-fold leap, per Anthropic’s red team notes (Anthropic Red Team). Mozilla got direct access outside Project Glasswing, Anthropic’s consortium of giants like Amazon, Microsoft, and JPMorgan racing to patch before wider release.

Holley sees a finite crisis ahead. “Every piece of software is going to have to make this transition, because every piece of software has a lot of bugs buried underneath the surface that are now discoverable,” he told Wired. Firefox rounded the curve first. Larger firms now pull thousands of engineers for six-month sprints. Smaller open-source projects? They’re drowning. Holley warns of maintainers lacking tools or bandwidth: “It’s difficult for these maintainers to not only have the wherewithal and the access to be able to use these tools, but also to actually do anything with them.”

Mozilla CTO Raffi Krikorian echoes the strain in a New York Times op-ed: “The underlying economics haven’t changed. The most valuable software infrastructure in the world continues to be maintained by people working for free, while the companies building fortunes on top of it never had to pay for its upkeep.” Mythos spares no one. It dug up a 27-year-old OpenBSD flaw and a 16-year-old FFmpeg bug that fuzzers missed five million times. Firefox felt the heat too.

So how does Mythos work? It reads code like a veteran exploit dev—mapping memory, grasping compilers, chaining flaws. No specific training. Capabilities emerged from scaling. In Firefox tests, it built JIT heap sprays, escaped sandboxes, and chained four bugs into full renderer takeovers. Wall Street Journal reported engineers begging Anthropic: “What else do you have? Send us more” after the initial Opus scan (WSJ). TechCrunch noted the pace: 22 vulns in weeks, fixed in 148 (TechCrunch).

The 271 bugs span memory issues, logic errors, use-after-frees—classes fuzzing often skips. Mozilla validated each, landed fixes. But open source lags. Abandonware sits exposed. Holley pushes informal ties across ecosystems: “Ultimately the open source stuff is a human problem. There’s only so much that you can scale with technology—there’s a lot of the industry and everybody just needing to come together.”

And attackers? They’ll get these tools. Anthropic and OpenAI limit releases, form working groups. Mozilla’s head start buys time. Firefox 150 ships protections now. Yet Krikorian warns of resource haves and have-nots. Big Tech fortifies. Volunteers scramble.

Recent X chatter underscores the buzz. Posts from April 21 highlight Wired’s update—from 151 to 271 bugs fixed. Developers eye the shift. One thread debates Mythos exploits in SpiderMonkey shells, sans full sandbox. Real-world impact? Patches landed anyway.

This isn’t hype. It’s triage. Software’s underbelly exposed. Firefox leads the fix. Others follow—or fall behind.

Subscribe for Updates

AISecurityPro Newsletter

A focused newsletter covering the security, risk, and governance challenges emerging from the rapid adoption of artificial intelligence.

By signing up for our newsletter you agree to receive content related to ientry.com / webpronews.com and our affiliate partners. For additional information refer to our terms of service.

Notice an error?

Help us improve our content by reporting any issues you find.

Get the WebProNews newsletter delivered to your inbox

Get the free daily newsletter read by decision makers

Subscribe
Advertise with Us

Ready to get started?

Get our media kit

Advertise with Us