A federal lawsuit alleges Google flipped a switch so its AI could read your emails. Google says that's misleading and denies it. Here's what's actually happening — the claims, the denial, the timeline, and the one-minute check you can run yourself.
Triggered by a viral post by Andy Diaz (Oct 15, 2026). Every legal claim below is an allegation, not a finding — the case is still in court.
ALLEGATIONThele v. Google LLC, Case No. 5:25-cv-09704, U.S. District Court for the Northern District of California (San Jose). A proposed class action filed November 11, 2025.
The plaintiffs allege that on or around October 10, 2025, Google changed its Gemini "Smart features and personalization" setting from opt-in to ON by default across Gmail, Google Chat, and Google Meet — letting Gemini access users' private communications (emails, attachments, chats, meetings) unless each user found the setting and turned it off themselves. The legal basis is the California Invasion of Privacy Act.
Advocacy pieces about the case cite a figure of 1.8 billion Gmail users potentially affected. That figure comes from the complaint and advocacy coverage — attribute it as claimed, not independently confirmed.
DENIALA Google spokesperson said the reports about the case are misleading, that no settings were changed, and that "we do not use your Gmail content to train our Gemini AI model."
Both sides are presented here because the case is unresolved. Allegations are not findings — read on for where the case stands.
There is no settlement, no certified class, and no claim form. If anyone contacts you claiming a payout from this case, that is a separate scam.
Whatever the court decides, you can control this setting today:
Verified against Gmail's settings as of October 2026.
These are the capabilities the setting controls:
The honest trade-off: to draft, summarize, or answer questions about your messages, the AI must access their content. That's the bargain being litigated — convenience powered by access.
At Google's "Gemini at Work" 2026 event (Oct 8), Google Cloud CEO Thomas Kurian announced the Gemini agent — enterprise coworker agents that get their own Workspace account:
something@agents.company.com).Important correction to the viral framing: this is enterprise Workspace in private preview (no pricing or general-availability date announced) — not consumer Gmail. It does not "live inside your inbox" as an ordinary Gmail user. If you ever see an unfamiliar agent address in your company's directory, ask IT before sharing anything with it.
Right now, the honest answer is: Google's systems, and — if the lawsuit's claims hold — its AI models too. Thele v. Google alleges Smart features were on by default so Gemini could read emails. Take the 5-question checkup to see what your own inbox exposes.
1. Do you know where Gmail's "Smart features and personalization" setting is?
2. Have you checked that setting in the last month?
3. Do you use a separate email address for banking and financial accounts?
4. If a coworker AI agent got its own email address in your company directory, would you know how to check whether it is legitimate?
5. Do you know that an allegation in a lawsuit is not the same as a finding by a court?
That is the allegation in an active federal lawsuit, not an established finding. The plaintiffs in Thele v. Google allege that Gmail's "Smart features and personalization" was switched on by default so Gemini could access private communications. Google denies this, saying no settings were changed and Gmail content is not used to train Gemini. The case is still live with no settlement or certified class.
It is the Gmail setting that controls whether Google's AI features can use your mailbox content to offer drafting help, email summaries, and plain-language inbox search. Because those features operate on your messages, they need access to them. To turn it off: Gmail, Settings, See all settings, General, then "Smart features and personalization", then Off.
No. The case (Thele v. Google LLC, Case No. 5:25-cv-09704, N.D. Cal.) is still live. The first complaint was dismissed on July 7, 2026; a second amended complaint was filed August 18, 2026; Google filed a second motion to dismiss on September 29, 2026. There is no settlement, no certified class, and no claim form — anyone telling you otherwise is scamming you.
Google denies the allegations. A spokesperson said reports about the case are misleading, that no settings were changed, and that Google does not use Gmail content to train its Gemini AI model.
At Google's "Gemini at Work" 2026 event, CEO Thomas Kurian announced enterprise coworker agents that get their own Workspace account — an email address, calendar, Drive storage, and a company directory seat — with actions logged under the agent's own identity. This is enterprise Workspace in private preview, not consumer Gmail: it does not "live inside your inbox" as a regular Gmail user.
FROM INGENIOUS AI — all 11 platforms: DisclosureGuard · ModelMinder · TermsWatch · Vibe-Code Audit · Rogue-Agent Defense · BreachHunter · ChatBreaker · Agent Flight Recorder · Agent Auth Gateway · DomainGuard · ContentCraft AI