Apple has temporarily restricted the number of bug reports researchers can submit because its review pipeline is overwhelmed by low-quality, AI-generated entries containing hallucinated vulnerabilities. This administrative clog prevented Italian startup Bynario from reporting a serious macOS flaw that grants attackers full control of a machine. The company’s CEO, Alfredo Pesoli, estimates the unreported vulnerability is worth between $100,000 and $200,000 on the black market. Apple has since contacted Bynario to address the issue directly.
The situation highlights a growing tension where artificial intelligence hinders human security efforts while simultaneously being deployed to find flaws internally. Apple is currently using tools from Anthropic and OpenAI to hunt for vulnerabilities, a strategy that has increased the number of fixes in recent updates by five times. Experts warn that bug bounty programmes may struggle to survive if they cannot filter out machine-generated noise at scale. Rafe Pilling of Sophos noted that these programmes have shifted from finding new issues to validating them at machine speed.
* The restriction was imposed after the inbox flooded with submissions lacking genuine technical merit.
* Internal AI tools at Apple have multiplied the volume of discovered security fixes significantly.
* Industry analysts question whether crowdsourced discovery models can function when automated content dominates the input stream.
