Instagram Users Hit by AI Chatbot Exploit
Instagram is sending notifications to users whose accounts were targeted by hackers who weaponized Meta's AI-powered support chatbot — and the attacks continued even after Meta claimed the issue had been resolved.
The breach came to light after reports emerged that bad actors had found a way to abuse Meta's AI support assistant to gain unauthorized access to Instagram accounts. The chatbot, designed to help users recover accounts and troubleshoot issues, was instead being manipulated into handing over access to hackers posing as legitimate account owners.
How the Attack Worked
While Meta has not disclosed the full technical details, the attack appears to have exploited the chatbot's account recovery and verification flows. Hackers reportedly manipulated the AI system into bypassing standard security checks, effectively tricking it into granting them control of victims' accounts.
What made the incident particularly alarming was that accounts continued to be compromised even after Meta announced it had patched the vulnerability. That gap between the company's assurances and the ongoing attacks has left many users — and security researchers — questioning how thoroughly the fix was tested before rollout.
Meta's Response
Meta acknowledged the issue and began notifying affected users, though the company has faced criticism for the delay in alerting victims and for its initial claim that the problem had been resolved when it apparently had not.
The notifications being sent to users mark an attempt at transparency, but many in the cybersecurity community argue that disclosure should have come sooner — and that the root problem points to a broader structural issue: AI chatbots handling sensitive account recovery functions are an attractive and potentially underprotected target for attackers.
A Growing Security Concern
This incident fits into a widening pattern of AI systems being exploited in unexpected ways. As companies rush to deploy AI assistants across customer-facing services — from support chats to account management — security researchers have repeatedly warned that these systems introduce new attack surfaces that traditional security testing doesn't always catch.
Large language models can be vulnerable to prompt injection attacks, social engineering, and edge cases that fool them into taking actions they shouldn't. When those systems have the ability to modify account credentials or grant access, the stakes are significantly higher.
What Affected Users Should Do
If you received a notification from Instagram about this incident, security experts recommend taking the following steps:
- Change your password immediately and choose something unique to your Instagram account
- Enable two-factor authentication if you haven't already — preferably using an authenticator app rather than SMS
- Review connected third-party apps in your account settings and revoke access to anything you don't recognize
- Check your account activity for any posts, messages, or setting changes you didn't make
- Log out of all sessions using the option in your account security settings
For users who weren't notified but are concerned, reviewing your security settings and enabling two-factor authentication is always a good precaution.
The Bigger Picture
This incident serves as a cautionary tale for the tech industry at large. Deploying AI in high-stakes, security-sensitive contexts requires a different level of scrutiny than rolling out a product recommendation engine or a chatbot that answers FAQ questions.
As AI becomes more deeply embedded in the infrastructure of platforms billions of people rely on, the consequences of getting security wrong grow proportionally larger.
Source: TechCrunch
