Meta announced two advertising changes within two days: an AI-powered overhaul of ad review (October 7) and a restriction period for US social-issue, election and political ads running October 27 – November 3, 2026 (October 6).
What happened
AI ad review (Oct 7). Meta's official blog describes large language models that detect "signposting" — seemingly benign ads that covertly steer people to illegal content off-platform. Review now also covers where an ad leads, not only what it shows: landing pages are scanned too. Meta is additionally running AI sweeps over existing ads, deploying a red-teaming agent to probe its own enforcement, and strengthening detection of repeat offenders who reopen accounts after being banned.
US election ad freeze (Oct 6). In a separate developer-blog post, Meta announced that from 12:01 AM PT on October 27 through 11:59 PM PT on November 3, 2026, advertisers may not create new ads about social issues, elections or politics in the United States. Ads that had a valid disclaimer and at least one impression before the cutoff may keep running, but only limited edits (bid, budget, end date, pause) are allowed.
Who's affected
Advertisers on Meta face a higher false-positive rate while the LLM rollout beds in. Anyone planning US political or social-issue campaigns gets a 7-day blackout on new ads. And anyone relying on "reopen a new account after a ban" should expect repeat-offender detection to make that materially harder.
KappS's take
- US political advertisers: get your ads live with disclaimers — and at least one impression — before October 27. After that, no new launches for a week.
- Expect more LLM misfires: early AI review waves historically over-block. Wrongful rejections are among the easiest appeal wins — document everything.
- Stop counting on reopens: Meta is explicitly targeting ban-evading reopens. Factor a lower reopen success rate into your risk planning.