AI Floods Apple With Bug Reports Faster Than It Can Vet

AI Turns Bug Hunting Into a Firehose

Apple has begun capping how many vulnerability reports it will accept from researchers at once, a response to a surge of AI-generated submissions that its security team can no longer triage at normal speed. Some of these AI-assisted reports are turning out to be genuine, patch-worthy Mac vulnerabilities, but many others are low-quality or speculative, forcing Apple to throttle intake even as real threats slip into the same queue 1.

The bottleneck isn't unique to Apple. Across the industry, AI-driven vulnerability discovery is producing findings at a volume traditional review processes were never built to handle. One tally puts the number of newly reported software flaws tied to AI-assisted scanning above 45,000, a figure that underscores both the promise of automated discovery and the strain it places on enterprise patching cycles — and the risk that the same tools could be repurposed for offensive exploitation 4.

The Big Players Are Racing to Build Better Hunters

Google has emerged as the most visible example of AI vulnerability-hunting done at scale. The company says it has used AI systems to patch 1,072 vulnerabilities in Chrome and now plans to ship security updates weekly rather than on its previous slower cadence 5. Separately, Google's AI agent harness has been credited with uncovering a Chrome flaw that had lingered undetected in the codebase for 13 years, a discovery framed as evidence of both the technology's power and how much latent risk has been sitting unpatched in widely used software 7.

Other vendors are building competing or complementary tools. Cisco has launched its Antares family of AI models specifically to help security teams parse unfamiliar codebases, manage the naming complexity of tracking vulnerabilities across projects, and trace how flaws propagate through dependencies 2. Meanwhile, Wiz and Google have introduced a multi-model system positioned as an answer to a system called Mythos, claiming better performance at surfacing software vulnerabilities and offering defenders a new architecture to work from rather than relying on a single model 3.

Policy Hasn't Caught Up

The rapid rise of AI-assisted hacking and defense has outpaced regulation. Following a reported AI-related hack, a policy expert has called for greater transparency, warning that the absence of clear rules around AI's offensive and defensive capabilities is leaving gaps in security oversight. That warning came the same week OpenAI CEO Sam Altman was on Capitol Hill discussing how Washington should approach governing increasingly capable AI systems 6.

Taken together, the coverage describes an inflection point: AI is genuinely accelerating vulnerability discovery — sometimes finding flaws humans missed for over a decade — but it is also overwhelming the human review pipelines meant to validate and act on those findings, raising questions about how vendors and regulators keep pace.