Report says Iran is using ChatGPT and Gemini to boost cyber attacks

In short: Western AI chat tools like ChatGPT and Google Gemini are being used by Iran-linked hackers to make cyber attacks faster and more convincing.

What's going on

A Financial Times report says Iran’s cyber groups are leaning on Western AI models, including ChatGPT and Google’s Gemini, to support hacking work. Cyber security experts quoted in the report say they are seeing signs that AI is being used throughout the process.

The reported uses include writing malware (harmful software, like a digital break-in tool), scanning the internet for weak spots in systems, and creating phishing messages. Phishing is when attackers pretend to be someone trustworthy to trick you into clicking a bad link or sharing passwords, like a fake bank email. The report says AI can also help these groups write messages in fluent Hebrew and Arabic, and even maintain believable fake online identities over long conversations.

The report points to recent examples in the region. The United Arab Emirates has said it faced more than half a million cyber attacks per day during the fighting, and that some were assisted by ChatGPT. The report also says Israelis received waves of phishing texts and emails, including messages that reportedly tried to recruit people to share intelligence.

What to watch

Tech companies say they try to block misuse, but the report describes the problem as “whack-a-mole,” where new accounts pop up after old ones are removed. OpenAI said it disables accounts tied to harmful activity and that its safeguards do not provide “novel cyber capabilities,” meaning nothing completely new that hackers could not do before. Watch for more public reports from OpenAI, Google, and security firms on how often these tools are being abused, and what restrictions, if any, get tightened.

Source: Financial Times