Can border agents see your ChatGPT? What a US border phone search reaches in your AI chats
The Morse teamPublished 6 min read
As of October 2026, US border agents can see your ChatGPT chats if the app has them on your phone or laptop when you cross, but CBP's own rules bar officers from using your device to pull up chats stored only on OpenAI's servers. Customs and Border Protection officers search what is "resident upon the device," with the phone in airplane mode, so the answer turns on where each AI app keeps your history.
We read CBP Directive 3340-049B, which replaced the 2018 directive on January 1, 2026, CBP's border-search FAQ and quarterly numbers, the tear sheet officers hand travelers, two federal appeals court opinions, and OpenAI's, Anthropic's, and Google's help pages on October 23, 2026. The new directive keeps the 2018 version's rule on remote data and adds Bluetooth and Wi-Fi to what officers switch off. This isn't legal advice.
How often does CBP search phones at the border?
Rarely, though CBP's own figures show it rising. In fiscal year 2025 (October 2024 to September 2025), CBP processed 419,157,270 travelers and searched the devices of 55,318: 41,728 non-citizens and 13,590 US citizens (CBP FY2025 statistics).
Fiscal 2026 is on pace to pass it. CBP's quarterly table, last updated August 4, 2026, shows 52,081 searches from October 2025 to June 2026, 94% of fiscal 2025's total in three quarters. Advanced searches, where an officer plugs in equipment to copy the phone, reached 4,564, already more than fiscal 2025's 4,396.
What can CBP look at on your phone?
Anything stored on it, in one of two kinds of search (CBP Directive 3340-049B):
- A basic search is an officer going through the device by hand, "with or without suspicion."
- An advanced search is one where an officer "connects equipment, wired or wireless, to copy and/or analyze the contents." It needs reasonable suspicion or a national security concern, and a Grade 14 supervisor's approval. New in 2026, one based on national security alone also needs a senior official such as a Director of Field Operations to sign off.
Using equipment "merely" to "bypass a password, overcome encryption, translate content" doesn't make a search advanced.
Can border agents see chats stored in the cloud?
Not under CBP's policy. The directive says officers "may not intentionally use the device to access information that is solely stored remotely," and that they will ask you to disconnect "by placing the device in airplane mode and disabling Bluetooth and Wi-Fi connections," or do it themselves. Passcodes you give "may not be utilized to access information that is only stored remotely." But the search does cover what's "accessible through the device's operating system or through other software, tools, or applications," which includes an app's stored data.
That rule is policy, and the directive says it "does not create or confer any rights." Courts haven't settled it. In United States v. Cano, the Ninth Circuit said it "need not address" data a phone "gives access to, but does not contain in its own memory" (Ninth Circuit opinion).
What do ChatGPT, Claude, and Gemini keep on your device?
Some, but the companies don't say how much. We found no page from OpenAI, Anthropic, or Google listing what their phone apps keep for offline reading. Their own pages say this much:
- ChatGPT on a phone. History "syncs across devices where you're signed in" (OpenAI). OpenAI's locked chats page says that on iOS, "previously loaded chat content may also remain visible until the app refreshes" (OpenAI), so the app holds some content between refreshes.
- ChatGPT on a laptop. Local Work chats in the desktop app "run on your computer" (OpenAI).
- Claude Code. Anthropic is the most specific. Its clients "store session transcripts locally in plaintext under
~/.claude/projects/for 30 days by default," and sessions last used in Claude Desktop or Cowork are exempt from that limit (Anthropic). - Gemini. Google keeps chats with your Google Account, for 72 hours even with Keep Activity off (Google), and doesn't say what the app caches.
Exports, screenshots, and downloads are ordinary files on the device. For what each company keeps on its servers, see our comparison of what the AI apps store.
What does each kind of border search reach in your AI chats?
We matched CBP's directive and FAQ to where AI chats live, as of October 23, 2026.
| Where the AI chat is | Basic search (no suspicion needed) | Advanced search (suspicion or national security concern) | Chats stored only on the company's servers |
|---|---|---|---|
| A signed-in AI app on your phone | What the app shows offline, which depends on its cache | The app's stored data, copied | Out of scope: officers may not use the device to reach them |
| Older history the app never loaded | Not on the device | Not on the device | Out of scope |
| Transcripts, exports, and screenshots | Opened like any file, including Claude Code's plaintext transcripts | Copied with the drive | Not applicable: they're local |
| A locked or encrypted app | The device can be detained; a tool that only bypasses the lock still counts as basic | Copied, then opened if CBP gets past the encryption | Out of scope |
| What CBP keeps | Notes on what was seen | Copies for up to 21 days, longer with probable cause | Reachable by a warrant to the AI company instead |
A chat beyond a border search's reach still sits on the company's servers, where a warrant can reach it any day of the year, as we cover in can AI chats be subpoenaed.
Do you have to unlock your phone for CBP?
CBP says travelers are obligated to present devices "in a condition that allows for the examination of the device and its contents" (CBP), and states the consequences:
- US citizens "will not be denied entry," but the device "may be subject to exclusion, detention, or other appropriate action."
- Foreign nationals may have their noncompliance weighed in the admissibility decision, though CBP says its ability to inspect a device "alone does not determine admissibility."
CBP's tear sheet says you "can always ask to speak with a CBP supervisor" (CBP tear sheet). Never lie to an officer, and ask a lawyer before you fly if you're unsure. Visitors face a separate question. A December 10, 2025 notice proposed making five years of social media mandatory on the ESTA application (Federal Register), and we found no notice finalizing it as of October 23, 2026.
Do courts require a warrant for a border phone search?
Not in the two appeals courts we read, though they split on scope. Cano (Ninth Circuit, August 2019) required reasonable suspicion for forensic searches and limited all phone searches to looking for "digital contraband." Alasaad v. Mayorkas (First Circuit, February 2021) held that "neither a warrant nor probable cause is required," and said of Cano, "We cannot agree with its narrow view" (First Circuit opinion). CBP's directive notes that "different requirements may apply in certain jurisdictions."
How Morse keeps your AI history off servers and locked on the device
Most AI chats exist twice, as a copy on the company's servers that legal process can reach any time and as whatever the app leaves on your device, in a form the company doesn't fully describe. Morse is built with no server copy. Our servers don't keep messages or log message content.
Your history lives in your browser, stored encrypted and locked when you sign out. Unlocking it takes two keys, one that never leaves your device and one our server gives you when you sign in, so a copy of your browser data alone can't be read. It is still on your device, and we won't guess how a border search would treat it.
Each chat also picks who handles the message on its way:
- E2EE, for anything no one but you should read. Your browser checks the secure enclave's proof, then encrypts the message so only that enclave can read it, and our servers relay text they can't read. The enclaves are run by third-party providers, not by Morse, and the guarantee depends on the enclave hardware and attestation working as designed.
- Private, for trip planning and everyday questions. It's the default for a new chat, and Morse has zero data retention agreements with its inference, search, and other providers.
- Anonymous, to ask GPT, Claude, or Gemini without the provider knowing it's you. The provider may keep the message for a limited time to check for abuse, under its own policy.
In every mode, Morse doesn't train on your chats, sell your data, or show ads, and because we don't store chats, we usually have very little to give authorities (privacy policy). A Temporary chat is kept in memory only and never saved. Clearing your browser or losing the device deletes your history for good, so save anything you need. More in AI that doesn't save your data.
Try a chat from the Morse home page without an account, or start free with no card needed. End-to-end encrypted chats come with Pro, at $19 a month.
Common questions
How long can CBP keep my phone?
CBP's directive says a detention for a search "ordinarily should not exceed five (5) calendar days," with longer extensions approved by senior managers. You get a written custody receipt, Form 6051D, naming a contact at CBP.
Does CBP keep a copy of what it finds in my AI chats?
It may keep copies for up to 21 days while it decides, and longer if it finds probable cause or the information relates to immigration, customs, or other enforcement matters. CBP says retained information may sit in its Automated Targeting System for up to 15 years.
What if my AI chats include messages to or from my lawyer?
Tell the officer. The directive says an officer who meets material asserted to be privileged must ask which files it covers and contact CBP's counsel's office before searching them, and a filter team or another measure sets it apart. Ask your lawyer before you travel with case files.
Do the same rules apply when I leave the United States?
Yes. The directive covers inbound and outbound border searches. Homeland Security Investigations agents can also search devices at the border under ICE's own policy, which this directive doesn't govern.