On November 17, 2026, Forbes reported that OpenAI and cybersecurity company Calif had used GPT-6 Cyber over recent months to identify about 35 vulnerabilities across Apple hardware and software. Some flaws could let an attacker take over an iPhone or Mac through a malicious website, the report said; researchers also reportedly coded an exploit that removed some protections in Lockdown Mode. Read the report.
The report named iPhone, Mac, iPad, Apple Watch, Apple TV and Apple Vision Pro among the product families covered by the research. NeoTeo previously covered OpenAI’s reported GPT-6 Cyber findings in Firefox, a separate account of the model’s use in security research.
What the reported findings could mean
The reported malicious-website scenario concerned potential control of an iPhone or Mac. Separately, researchers reportedly coded an exploit that removed some Lockdown Mode protections. Calif CEO Thai Duong said human researchers still need to provide the methodology and verify findings. He continued to recommend Lockdown Mode for protection against a wide range of attacks.
AI can help surface a security lead; human researchers still need to supply the method and verify the finding.
What Calif and OpenAI said about disclosure
Calif told Forbes that Apple had fixed many of the reported issues after receiving alerts. OpenAI told the publication that Apple had credited GPT-6 Cyber for vulnerability findings.
In a separate post, Calif said it had reported 36 vulnerabilities to Apple “since March.” Calif’s post described the company’s Apple disclosures.