OpenAI a découvert deux failles critiques dans PHP grâce à ses modèles d’intelligence artificielle, dans le cadre de l’initiative Patch the Planet. According to Greg Brockman, the bugs had been present for 14 years, while versions were discovered by Calif in August and by OpenAI in September, and Wiz also found the flaws independently. The vulnerabilities allowed attackers to execute code on PHP servers, which power around 70% of websites with a known server-side language. Following the disclosure, Finland, the Netherlands and France temporarily took government websites offline, while PHP released a fix one day after the details began to circulate. Brockman also proposed that authorized defenders, including governments, could scan for vulnerable servers and remotely patch them in extreme cases, with OpenAI committing $1 billion in subsidized access to its Daybreak program for critical-infrastructure defenders.
Source: Lire l’article original
