GPT-4 capable of autonomously exploiting vulnerabilities in real world systems through security advisories, says study


by MediaNama

MediaNama— In a sandbox environment, GPT-4 was capable of autonomously exploiting 87% of real-world vulnerabilities , which would come out to 13 out of 15, by reading Common Vulnerabilities and Exposures (CVE) advisories.

Mashable—ChatGPT Plus can exploit zero-day security vulnerabilities — why this should concern you. GPT-4 can exploit zero-day security vulnerabilities

Tuoi Tre News Video—Real Madrid secure LaLiga title after Girona thrash Barcelona 4-2. The result left Carlo Ancelotti's side with an unassailable lead in the standings

Phys.org—Study says it's likely a warmer world made deadly Dubai downpours heavier. Circumstantial evidence points to climate change as worsening the deadly deluge that just flooded Dubai and other parts of the Persian Gulf, but scientists didn't discover the definitive fingerprints of greenhouse gas-triggered warming they have seen in other extreme weather events, a new report found.