Meta's AI Tools Are Creating New Security and Privacy Risks for Instagram Users
A new AI image tool and a hijacked support bot show how Meta's AI features can be turned against ordinary accounts.
What matters
- CNET reported that Meta's new AI image tool (tied to 'Muse' AI image/content-reuse capabilities) can be used to create deepfake content of Instagram users, and recommends opting out of a relevant AI setting.
- Cybersecurity researchers found attackers exploited Meta's AI support assistant to hijack high-profile Instagram accounts, including the former White House account (2.4M+ followers), a senior U.S. Space Force member, and Sephora.
- Attackers prompted Meta's AI chatbot to trigger password resets and used AI-generated selfie videos to pass identity verification.
- Meta says the support-bot vulnerability has been fixed and impacted accounts are being secured.
- Both incidents highlight how Meta's AI features create new attack surfaces for impersonation and account takeover.
What happened
Two separate but related incidents highlight how Meta's AI-powered features on Instagram are creating new attack surfaces.
First, a CNET reporter documented how Meta's new AI image tool—referenced in the publication's reporting as tied to the company's "Muse" AI image and content-reuse capabilities—could be used to create convincing deepfake content of a friend's Instagram. The reporter's headline-level finding: Instagram users may want to opt out of an AI setting that appears to enable this kind of content reuse, and the experience underscored how easily the tool could be misused. Full details from CNET's report were limited at the time of this article.
Second, and more alarmingly, cybersecurity researchers found that attackers exploited a vulnerability in Meta's AI-powered support assistant to take over several high-profile Instagram accounts. According to reporting by CRBC News, affected accounts included the former White House Instagram account used during the Obama administration (with over 2.4 million followers), a senior member of the U.S. Space Force, and the beauty brand Sephora.
The attack worked by prompting Meta's AI chatbot to initiate password resets for targeted accounts and to send verification codes to email addresses controlled by the attackers. When the bot requested identity verification via a selfie video, the attackers allegedly presented AI-generated fake selfie videos to pass the check. They then changed the account recovery email and reset the password. Videos demonstrating the attack were reportedly shared in Telegram groups for security researchers.
Meta has stated that the issue has been fixed and that impacted accounts are being secured.
Why it matters
These incidents illustrate a broader pattern: as Meta layers AI features into Instagram—from image generation to automated customer support—the platform is simultaneously expanding the toolkit available to legitimate users and to bad actors. The support-bot exploit is particularly concerning because it targeted accounts with millions of followers and relied on Meta's own AI infrastructure to bypass the company's identity-verification safeguards.
The CNET report on the image tool points to a different but parallel risk: AI features that reuse or repurpose user content can blur the line between creative tools and impersonation enablers. If users are opted in by default to settings that allow their likenesses or content to be used in AI-generated outputs, the potential for misuse grows—especially when combined with the kind of deepfake selfie videos that researchers say were used in the account-takeover attacks.
Together, these stories suggest that Meta's AI rollouts are outpacing the guardrails needed to keep users safe.
Public reaction
No strong public signal was available from Reddit or other discussion platforms at the time of this article. The story is still developing, and broader community reaction may emerge as more details from the CNET report and the security research become public.
What to watch
- Whether Meta provides more detail on the AI image tool's content-reuse settings and whether opt-out is truly granular or buried in menus.
- Further technical analysis of the AI support-bot exploit, including whether the selfie-video verification has been meaningfully strengthened.
- Any policy changes from Meta regarding default opt-in for AI training or content-reuse features on Instagram.
- Whether affected high-profile account holders publicly comment on the hijackings.
Sources
Public reaction
No Reddit or public discussion data was available at the time of this article. Community reaction may emerge as the CNET report's full details and the security research become more widely circulated.
Open questions
- How widespread was the AI support-bot exploit before Meta patched it?
- Will Meta change default opt-in settings for AI content reuse on Instagram?
- Are ordinary users able to easily find and disable the AI setting flagged by CNET?
What to do next
Developers
Review Meta's AI API and platform documentation for any content-reuse or image-generation endpoints that could be abused for impersonation, and build detection safeguards if integrating Meta AI features.
The support-bot exploit shows that AI-powered verification flows can be bypassed with synthetic media; developers building on Meta's stack need to understand these failure modes.
Founders
Audit your startup's social account recovery and identity-verification flows for AI-bypass vulnerabilities, especially if you rely on selfie-video checks.
The Instagram hijackings demonstrate that AI-generated selfie videos can defeat automated identity checks, a risk any founder managing high-value accounts should address.
PMs
Evaluate whether your product's AI features have clear, accessible opt-out mechanisms for content reuse, and ensure default settings prioritize user consent.
The CNET report highlights user frustration with AI settings that may be enabled by default; PMs should ensure transparency and control to avoid trust erosion.
Investors
Monitor Meta's response to these AI security incidents for signs of regulatory scrutiny or user churn that could affect platform engagement metrics.
Repeated AI-driven security failures on a platform with billions of users could attract regulatory action and damage advertiser confidence.
Operators
Immediately review and tighten Instagram account security for your organization's official accounts: enable two-factor authentication, audit recovery email settings, and check AI-related privacy opt-outs.
High-profile accounts including brands like Sephora were targeted in the support-bot exploit; operators managing brand accounts should assume they are potential targets.
How to test
- 1Open Instagram settings and navigate to the account privacy or AI settings section.
- 2Look for any setting related to AI content reuse, AI training, or image generation permissions.
- 3If such a setting exists and is enabled by default, toggle it off and note the granularity of the control.
- 4Attempt to access Meta's AI support assistant and observe what identity-verification steps it requires (do not attempt to exploit—simply document the flow).
Caveats
- The CNET report's full details were not available at time of writing; the exact setting name and location may differ from what is described here.
- Do not attempt to replicate the account-takeover exploit described in the CRBC News report—doing so would violate Meta's terms of service and potentially the law.
- Meta has stated the support-bot vulnerability is fixed, so the attack flow may no longer be reproducible.