Researchers warn threat actors are exploiting LLM hallucinations by registering fabricated domains that large language models sometimes invent, then using those sites for phishing and credential theft.
Attackers register domains that LLMs hallucinate and use them to host phishing pages.
Security researcher Seth Michael Larson calls the tactic “slop squatting” and urges users not to trust AI output without verification.
Large language models (LLMs) sometimes fabricate information, including web domains, when answering users' questions. Attackers are registering these hallucinated web domains to host phishing pages. Security researcher Seth Michael Larson, who dubbed this technique "slop squatting," told IEEE Spectrum that these attacks exploit the trust users place in AI tools.
Platform provides attack simulation and training, email and collaboration security, and agent security powered by AIDA (Artificial Intelligence Defense Agents) and a proprietary Risk Score. The platform leverages 15 years of behavioral data to combat advanced threats including social engineering, prompt injection, and shadow AI. ABOUT KNOWBE4 TEAM The KnowBe4 Team delivers timely, expert-driven insights on cybersecurity trends, emerging threat intelligence, human risk and agent security best practices, compliance strategies and industry research to help organizations strengthen their digital defense layer and stay informed, resilient, and secure.
They'll assume some security trust decisions have happened. You should not be trusting this," Larson says. "As in any engineering field, you can't just go off of vibes.
Malwarebytes reports that the scam, which appears as a website offer from Anthropic, is not as easy to detect as most.
Fake storefronts appeared days after the launch of Jev, an artificial intelligence model that returns decisions rather than written answers. The sites offer access to the service, but they put an unrelated operator between users and the official API. That means prompts and any information inside them travel through a t
Threat actors are abusing trusted AI platforms to host malicious content, poison search results, and trick users into installing malware. Huntress examines campaigns targeting AI users through weaponized Claude Artifacts, shared AI conversations, sponsored search results, and ClickFix-style lures. [...]
Threat actors are impersonating OpenAI and ChatGPT in phishing emails, using fake subscription payment alerts to create urgency and trick users into updating their billing information. The emails use legitimate-looking branding and redirect victims to convincing fake ChatGPT login pages designed to steal account creden
Scammers are using AI to generate phishing pages that impersonate antivirus products, according to Malwarebytes.
AI tools are drastically improving the speed of the reconnaissance stage of targeted social engineering attacks, according to researchers at ESET. Attackers can use these tools to trawl the internet for publicly available information about potential victims, and incorporate this information into personalized spear phis
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.