Industry

Google extends Gemini chat retention to 60 days, decouples history from training consent

Google is retaining temporary Gemini conversations for approximately two months instead of 72 hours, citing safety improvements and introducing granular privacy controls.

Last verified:

Google is fundamentally reshaping how it handles temporary Gemini conversations, extending retention from the previous 72-hour window to approximately two months—a fourfold increase—while simultaneously introducing granular privacy controls that separate chat storage from AI training consent. According to TechCrunch AI, the policy shifts globally over the coming weeks, representing a material change in how the company balances user experience with data stewardship.

Retention Period Doubles Industry Baseline

The two-month retention period marks a significant departure from Google’s prior practice. TechCrunch AI reports that archived versions of Google’s Gemini Apps Privacy Hub from November 12 documented the shift away from 72-hour retention. This aligns Google’s default behavior closer to industry norms: OpenAI maintains ChatGPT temporary chats for up to 30 days, while Anthropic retains Claude incognito sessions for 30 days by default. Google’s decision to extend retention by roughly 60 days suggests the company is prioritizing system reliability and incident investigation over minimizing data footprint.

Historically, Google tied chat retention and model training under a single “Keep Activity” toggle. TechCrunch AI reports that users now have independent control: they can preserve conversations in chat history without authorizing Google to use those chats for generative AI training. For users who previously disabled activity tracking, both settings default to off. Those who had the setting enabled will retain chat history by default, though they can opt out of training use independently. This architecture gives users finer-grained agency, though it complicates the privacy mental model.

Safety and System Resilience as Stated Justification

Google has not disclosed what prompted the retention extension, according to TechCrunch AI. Instead, the company cited a blog post attributing the change to improved safety response, user feedback processing, service protection, and system failure recovery. The updated Privacy Hub adds potential system failures as a retention rationale—suggesting that longer retention enables faster incident diagnosis. Google explicitly states that temporary conversations are not used for model training, a claim that contradicts the now-separate training consent toggle, raising a question about the temporal scope of that statement.

Why This Matters

The retention extension and consent separation carry upstream consequences for data-minimization advocates and emerging privacy regulation. If 60-day retention becomes the assumed baseline, regulators scrutinizing data-retention practices may feel pressure to endorse extended windows as “industry standard.” Conversely, the decoupled consent mechanism creates a template: users can now audit whether Google’s training use is proportionate to their actual consent, which opens visibility into model-training data sourcing.

For teams deploying Gemini in regulated environments (healthcare, finance, law), the two-month window may now require documented data-handling agreements. The phased rollout timeline remains vague—“over the coming weeks”—which suggests staggered enforcement and potential regional variation under GDPR or similar regimes.

Frequently Asked Questions

Will my Gemini chats be used to train Google's AI models if I disable history?

No. Google is decoupling chat storage from model training. Users can now keep conversations without consenting to training data use, and both settings default to off for users who previously disabled activity tracking.

How does Google's retention policy compare to competitors?

Google's 60-day retention is roughly double OpenAI's 30-day ChatGPT policy and Anthropic's 30-day Claude incognito default. However, all three companies reserve the right to retain data longer for legal, safety, or policy-violation investigation purposes.

When will this change take effect?

According to TechCrunch, the update will roll out globally in the Gemini app over the coming weeks, with the Privacy Hub already reflecting the new policy as of November 12.

#privacy #gemini #data-retention #ai-safety #google