Mozilla says Anthropic's Mythos Preview and other AI models helped it identify and ship 423 Firefox security bug fixes in April, compared to 31 a year earlier
When Anthropic unveiled its new Mythos model in April, it also delivered a stern warning to anyone developing software.
TechCrunchRussell Brandom
Context & Ripple Effects
Mozilla had already tied Firefox 150’s 271 vulnerability fixes to early access to Anthropic’s Mythos Preview. This update broadens that result: Mozilla attributes a much larger April remediation total to Mythos Preview alongside other AI models.
The result follows Anthropic’s launch of Project Glasswing, which positions Mythos Preview as a tool for finding and fixing software vulnerabilities, and Anthropic’s claim that the model uncovered serious issues across major software platforms.
First-order effects
Mozilla’s Firefox security team can move a substantially larger set of identified issues into shipped fixes, improving the browser’s near-term security posture.
Anthropic gains a concrete deployment reference for Mythos Preview in production vulnerability discovery and remediation workflows, rather than only model-reported findings.
Second-order effects
Browser and operating-system vendors face pressure to test comparable AI-assisted security workflows, particularly if model-aided discovery produces more actionable findings than existing review processes.
The bottleneck shifts toward validating, prioritizing, and safely patching a larger volume of AI-surfaced reports; security teams will need processes that distinguish exploitable bugs from noise.
Third-order effects
If similar results hold across vendors, AI models could make continuous code auditing a standard layer of software maintenance, raising both the expected pace of patching and the volume of vulnerabilities entering coordinated-disclosure pipelines.
The same capability creates a dual-use security dynamic: wider defensive adoption may be accompanied by demand from government and other sophisticated users seeking model-assisted discovery in widely deployed software.
The trend: This is part of the shift from AI as a coding assistant to AI as an active, high-throughput component of vulnerability discovery and remediation operations.
Nontechnical AI doomer folks latching onto this Mythos chart without putting in context the “all severities” and actual reality of the bugs. Typical. “Note that a number of these bugs are sandbox escapes, which would need to be combined with other exploits to achieve a [image]
Skepticism of corporate marketing and AI boosterism is always warranted, but I think the folks who accused Anthropic of overrating Mythos should check out this post by Mozilla developers indicating that the Firefox team fixed more security bugs in April using Mythos than in the […
So Mythos was, indeed, not marketing hype. Remember this is a general purpose model that just happens to be good at finding exploits because good models are good at lots of things. Expect similar from OpenAI & Google. And from open models in 8 months. https://hacks.mozilla.org/..…
One of the things that made the Mythos release hard to interpret is that Anthropic held back details on most vulns they found, to give defenders time to patch. 1 month later, info from orgs with access to Mythos is starting to trickle out, e.g. this post from Mozilla today: [imag…
Maybe powerful AI will favor the cyber defense. Vulnerabilities are finite in number, so if you can fix them all, maybe software just becomes much more secure.
At a moment when much of the industry is racing to use AI to extract value, we're focused on something simpler: using it to better protect people and the open internet. …
Maintaining Firefox used to require a scarecrow wearing a shirt that said “Mozilla” and stuffed full of Google's money but Anthropic managed to do it with Claude https://techcrunch.com/...
The stochastic parrot Claude Mythos Preview found 271 never-before-discovered bugs in Firefox, whose fixes revealed yet more bugs hacks.mozilla.org/2026/05/behi... [image]
A new post from Mozilla - the guys behind the Firefox browser, on what access to Anthropic's “Mythos” model for a few weeks had revealed in a 22-year old, open-source project. — Quite some detail, and also a striking chart. — hacks.mozilla.org/2026/05/behi... [image]
“We also continue to find bugs with other means internally, and, similar to other projects, we've seen a significant uptick in external reports in the last few months.” — hacks.mozilla.org/2026/05/behi...