Google Uses AI to Fix Chrome Bugs: But What Is the Other Side of the Medal?

Google has deployed AI-based systems to close security vulnerabilities in its popular Chrome browser. The company announced that thanks to artificial intelligence it fixed more bugs in June alone than in the previous two years combined. A standout success was closing critical flaws known as sandbox escapes, which let malicious software break out of the browser's protected sandbox and reach the computer system. To speed up the process, Google combined its own large language models with automated testing tools.However, the technology world is approaching this success story with scepticism. Many software experts stress that AI working alone on code often leads developers astray. For AI to work efficiently it must be fed telemetry—remote measurement data that reports how software performs back to central systems. AI models that are not fed the right data can produce slop code: structures that run but remain inefficient and overly complex. That raises the question of how many of the bugs cleared were reintroduced by AI in the first place.On the other hand, such success reports often reflect big tech companies' efforts to justify their AI investments. Engineers argue that firms share on…