Two people can sometimes remain signed into the same ChatGPT account at the same time, but OpenAI does not allow two different people to use one personal account as separate users. If you searched ‘can two people use the same ChatGPT account at the same time’, the decisive distinction is not whether two browser sessions can stay open; it is whether one person is using their own account on several devices or two people are sharing one identity. OpenAI explicitly permits the first and tells the second person to create their own account.
That distinction matters more in 2026 because a ChatGPT account is no longer just a password protecting a list of prompts. Depending on the plan and settings, it can hold conversation history, memory and personalisation, uploaded files, Projects, connected applications, billing information, custom GPTs, scheduled tasks and work context. When two people share one login, those systems do not become two private profiles. They remain attached to one account.
This guide separates what OpenAI actually documents from what third-party sharing guides merely infer. It also answers the harder practical questions: whether simultaneous sessions have a published limit, what happens to memory and history, whether two Plus subscriptions or ChatGPT Business make more sense for two people, how account switching differs from account sharing, and which collaboration methods avoid credential sharing altogether. Where OpenAI does not publish a numeric limit or behavioural guarantee, this article says so rather than inventing one. That evidence hierarchy matters because the live web is full of confident advice about IP addresses, device counts and supposedly safe workarounds that OpenAI itself has never documented.
The Rule OpenAI Actually Documents
OpenAI’s own Account Sharing Policy is unusually direct: an OpenAI account is meant for the individual who created it. If another person needs to use OpenAI products, the Help Centre tells that person to sign up for their own account. The policy also says the account owner may use the account on multiple devices. Those two statements solve most of the apparent contradiction around simultaneous use.
Think of the rule as an identity boundary rather than a device licence. A single person may move between a laptop, phone and tablet because the same identity is using the service. Two people taking turns—or typing at the same time—behind one personal login are different: the service can no longer reliably associate preferences, security events, limits and activity with one individual. OpenAI’s Terms of Use reinforce that point by prohibiting users from sharing credentials or making an account available to someone else.
The practical implication is that a successful login is not evidence that a usage pattern is permitted. A service may technically keep sessions alive across several devices while its terms still require one human account holder. That is why advice built around ‘it works on my two browsers’ answers the wrong question.
For a broader orientation to the product itself, our complete ChatGPT usage guide explains how modern ChatGPT accounts combine chat, files, research, memory and tools. The more of those capabilities a person uses, the more consequential a shared identity becomes.
There is one important uncertainty to preserve: OpenAI’s public Account Sharing Policy does not state a fixed maximum number of simultaneously active device sessions for one person. It says multiple devices are welcome and notes that usage limits may apply depending on account activity and subscription level. Therefore, any article that gives a universal ‘two devices’, ‘three devices’ or ‘five devices’ ceiling without an OpenAI source is presenting an inference as a rule.
| Situation | OpenAI-Documented Position | What It Means in Practice |
| One person, laptop + phone | Allowed to use the same account on multiple devices | Same user identity; plan and activity limits still apply. |
| Two people, one personal login | Account is intended for the individual who created it | Each person should have their own account. |
| Two co-workers in one Business workspace | Supported with separate paid seats | Shared administration without shared credentials. |
| One person, work + personal accounts | Supported through web account switching | Two accounts stay separate; only one is active at a time. |
Can Two People Use the Same ChatGPT Account at the Same Time?
The phrase ‘at the same time’ often pulls this question into a technical rabbit hole. OpenAI does not publish a consumer-facing concurrency specification that says one personal account can sustain exactly N simultaneous browser or app sessions. It does publish the higher-level rule that the same person may use the account on multiple devices. Those are not the same thing.
So what happens if a phone and a desktop are both signed in? For the account owner, that is an ordinary supported pattern. What happens if one person is on the phone while a partner, colleague or friend uses the desktop with the same credentials? The fact that both sessions may load does not turn the account into a two-seat subscription. OpenAI’s policy still treats the account as belonging to one individual.
Usage limits complicate the experience further. Plus and other personal subscriptions have plan-dependent limits that OpenAI says can vary with system conditions and account activity. If two people consume the same account’s allowance, they are drawing from one entitlement, not receiving independent quotas. A sudden limit can therefore appear to one person because of activity performed by the other. The same problem applies to billing, feature availability and usage history.
This is also why ‘same Wi-Fi’ is not a policy exception. OpenAI’s published account-sharing guidance does not say a spouse, sibling, roommate or colleague may share a personal login if both connect from one household IP address. Nor does it say that different IP addresses automatically prove prohibited sharing. Network location can be one security signal among many, but there is no official public threshold to cite. The correct rule is simpler: one personal account, one person. In other words, technical concurrency is an implementation detail that may change; named-user eligibility is the durable policy question readers should use when deciding how to set up access.
Why Shared AI Accounts Are Different From Shared Streaming Accounts
Streaming subscriptions trained people to think of sharing as a question of screens: if a service permits two concurrent streams, two people can watch. Generative AI accounts are structurally different. The valuable residue is not only what content was consumed; it is what the user asked, uploaded, reasoned through and asked the system to remember.
A peer-reviewed CHI 2026 study by Ji Eun Song, Eunchae Lee, Juhee Im, Hyunsoo Jang, Eunji Kim and Joongseek Lee examined shared large-language-model subscriptions through a survey of 245 users and interviews with 36 participants. The authors describe the distinctive privacy problem succinctly: shared LLM users can reveal ‘fragments of their cognition’. Their sample was exploratory and culturally situated rather than globally representative, so its percentages should not be universalised. But the mechanism they document is directly relevant: prompt histories can expose tentative thinking, emotional expression, strategies and unfinished reasoning.
That makes the observer effect unusually important. If you know another person can open the same history, you may stop asking candid questions about work, relationships, money, health, career doubts or creative ideas. The account can still function technically while becoming less useful precisely because people self-censor. Our wider analysis of 2026 AI privacy risks reaches a related conclusion: persistent context increases utility and exposure at the same time.
The research also helps explain why remote-desktop or anti-detect-browser workarounds miss the core issue. Hiding the second device, keeping both users on one machine, or avoiding a password hand-off may change the authentication path; it does not create two isolated personal contexts inside one account. The privacy and attribution problem remains because the service still sees one account identity. This distinction is the article’s core information gain: reducing authentication friction is not the same as creating privacy separation, independent personalisation or auditable user attribution.
“users reveal not just their usage patterns but also fragments of their cognition.” — Song et al., CHI 2026
What Gets Mixed When Two People Share One Login
The strongest argument against a shared personal login is not that another user will necessarily ‘break’ ChatGPT. It is that several layers of context can become entangled. OpenAI’s current account switcher gives a useful contrast: when one person keeps two accounts signed in on the web, OpenAI says chats, memory, history, billing and workspaces remain separate. A shared single account provides no equivalent identity-level partition between two humans.
Conversation history is the obvious layer. One person may see the other’s thread titles, messages, generated files or uploaded documents unless content is deleted, hidden or otherwise restricted by feature-specific controls. Memory and personalisation are subtler. ChatGPT can adapt to an account holder’s preferences and context; if two people deliberately feed conflicting preferences into one account, the resulting personalisation can be noisy or misleading. Our guide to how ChatGPT memory works explains why persistent context is valuable when it represents one person and harder to interpret when it represents several.
Files and Projects raise the stakes. A project may contain documents, instructions and conversations about one employer or client, while another person using the account may be working on unrelated material. Connected applications can extend that boundary beyond ChatGPT. OpenAI’s Business pricing page, for example, lists integrations with Google Workspace, Slack, GitHub and Microsoft 365 in the managed workspace product. Personal feature availability varies by plan and rollout, but the principle is stable: the more external context an account can reach, the more important it becomes to know whose identity authorised that access.
Billing and rate limits are shared too. If a personal Plus subscription belongs to one account, both people are effectively consuming the same plan’s limits. There is no built-in way to attribute one person’s usage, expenses or security events to a distinct seat. This becomes especially problematic when one user leaves a household, project or business relationship: changing a password cuts off a shared credential, but it does not retroactively separate histories, memories or files created under that identity.
| Account Layer | What a Shared Login Can Mix | Why Separate Identities Help |
| Chats and history | Prompts, responses, thread titles and generated work | Each person keeps a private conversation record. |
| Memory and personalisation | Preferences, recurring context and behavioural signals | Personalisation reflects one user rather than competing profiles. |
| Files and Projects | Uploads, instructions and project context | Work stays scoped to the correct owner or workspace. |
| Apps and connectors | Permissions may expose external work data through the active identity | Access can be governed per user and revoked cleanly. |
| Billing and limits | One subscription allowance and one payment identity | Costs and entitlements are easier to attribute. |
| Security and recovery | One password, one recovery path and one activity trail | MFA, passkeys and incident response map to an individual. |
The Two-Person Cost Equation in 2026
Cost is the reason many people consider sharing in the first place, so the answer should be numerical rather than moralistic. As of 22 September 2026, OpenAI lists ChatGPT Plus at $20 per month. ChatGPT Go is $8 per month in the United States, with pricing that can vary by market. Business Standard seats are $25 per user per month on monthly billing or $20 per user per month when billed annually. Business Premium seats are $125 monthly or $100 per month equivalent when billed annually. A Business workspace requires at least two paid seats.
For two Plus users, the public price is therefore $40 per month. For two Business Standard users, monthly billing is $50; annual billing is $480 upfront for the year, equivalent to $40 per month across two seats. The headline arithmetic makes annual Business Standard look identical to two Plus plans, but the products are designed for different contexts. Plus is an individual subscription. Business is an organisational workspace with central administration, SAML SSO and MFA support, usage analytics and spend controls, shared workspace capabilities, and no training on business data by default.
There is also a current 2026 wrinkle that older ranking pages miss. OpenAI added Premium Business seats in August 2026. Premium provides five times more usage than Standard and removes the five-hour usage limit. Business allows Standard and Premium seats to be mixed, and self-serve workspaces created under the current rules require at least two paid seats and support up to 200 paid Standard/Premium seats. Some older workspaces retain legacy conditions, so organisations should check their own billing screen.
Personal Pro is not a family or team licence either. OpenAI introduced $100 and $200 Pro tiers in 2026, but as of 10 September it has temporarily paused new sign-ups and upgrades to the $200 Pro tier. None of these personal tiers converts one account into multiple named users. For a detailed consumer-plan comparison, see our ChatGPT Free versus Plus guide, but always use OpenAI’s live checkout for local pricing and changing limits.
| Setup for Two People | Public US Price | Separate Identities? | Important 2026 Constraint |
| Two Free accounts | $0 | Yes | Usage and feature limits apply per account. |
| Two Go accounts | $16/month in the US | Yes | Go is an individual plan; local prices may differ. |
| Two Plus accounts | $40/month | Yes | Plus is $20 each; API usage is billed separately. |
| Business Standard, monthly | $50/month | Yes | Two-seat minimum; current self-serve cap is 200 paid seats. |
| Business Standard, annual | $480/year ($40/month equivalent) | Yes | Annual commitment; seat changes follow Business billing rules. |
| Business Premium, monthly | $250/month | Yes | Five times Standard usage and no five-hour usage limit. |
| Business Premium, annual | $2,400/year ($200/month equivalent) | Yes | Designed for high-usage organisational users. |
Which Setup Fits Couples, Families, and Teams
The right setup depends less on whether two people know and trust each other than on whether they need one shared identity. In most personal situations, they do not. A couple, siblings or roommates can each create a Free account and independently decide whether a paid tier is worth it. That preserves private histories and prevents one person’s use from consuming the other’s paid allowance. OpenAI does not currently publish a consumer family or duo plan that turns one personal subscription into two member profiles.
For two colleagues at the same organisation, the analysis changes. If the company needs central billing, managed access, shared workspace context or business-data protections, ChatGPT Business is the relevant self-serve plan. The fact that its minimum is two seats is important: OpenAI has designed the product so two named organisational users can exist inside one workspace without sharing a password. Our review of AI tools for business explains why identity, permissions and workflow governance increasingly matter as AI reaches company systems.
For a founder and assistant, manager and contractor, or agency and client, one login is especially poor architecture. The relationship may end, the scope of work may change, or one party may require access to only a subset of material. A named account or workspace member can be removed; a shared personal credential has to be rotated and leaves a muddled history of who did what.
A genuinely shared physical device is not the same as a shared ChatGPT account. Two people can use the same family computer while signing into their own accounts. On ChatGPT web, account switching is intended for one user who wants two accounts—typically personal and work—signed in at once. It can keep two accounts available in the same browser session while preserving separate chats, memory, files, billing and workspaces. That feature does not say two people should share one of those accounts.
| Scenario | Best-Supported Pattern | Reason |
| One person on phone, tablet and laptop | One personal account across devices | OpenAI explicitly allows multiple devices for the account owner. |
| Two family members | Two personal accounts | Keeps history, memory, limits and recovery separate. |
| Two colleagues in one company | Two Business seats where workspace controls are needed | Named users, central billing and organisational administration. |
| One person with work and personal identities | Account switching on ChatGPT web | Keeps two accounts signed in but independent. |
| Temporary contractor | Own account or managed workspace membership | Access can be scoped and revoked without sharing the owner’s credential. |
Supported Ways to Collaborate Without Sharing Credentials
People often share a login because they are really trying to share an outcome: a research thread, a file, a set of instructions, a reusable GPT, or access to company knowledge. Solving that underlying collaboration need usually removes the reason to share credentials.
For consumer collaboration, keep separate accounts and move the artefact rather than the identity. A user can share permitted conversation links or exported material, exchange source documents, or copy a prompt template into a shared document. OpenAI also added Library sharing in September 2026, allowing users to share files and folders with chosen people where the feature is available. Feature scope changes quickly, so verify the controls shown in the product before sharing sensitive material.
For research, the safest workflow is to maintain separate user accounts while sharing the evidence set and final synthesis. Our guide to research safely with ChatGPT recommends separating planning, retrieval, extraction, synthesis and audit. That workflow transfers well to teams because each person can preserve their own account history while collaborating on documents and sources.
For businesses, invite named members to a Business workspace rather than circulating the owner’s password. Assign Standard or Premium seats according to workload, configure SSO/MFA where appropriate, connect approved systems, and use workspace controls for shared context. Business pricing documentation says both seat types include ChatGPT, ChatGPT Work and Codex; Premium provides more included usage. API access is a separate product and billing surface, so sharing a ChatGPT subscription should never be treated as a substitute for programmatic multi-user access.
For one person who simply wants to separate contexts, use account switching rather than creating browser-profile workarounds. OpenAI’s current web switcher supports two signed-in accounts in one session, with only one active at a time. The separation is the feature: chats, memory, files, billing and workspaces do not merge.
Security, Recovery, and Connected-App Risk
A shared credential creates a security problem before any dramatic breach occurs: it weakens attribution. When one account produces a sensitive export, connects an app, changes a setting or triggers a security alert, the activity trail represents the account—not two independently authenticated humans. The owner is also responsible for activity under the account under OpenAI’s Terms.
OpenAI’s 2026 security work reflects how much context modern accounts can accumulate. In April, the company introduced Advanced Account Security protections aimed at reducing account-takeover risk for users whose ChatGPT accounts may contain sensitive personal or professional context and connected workflows. The lesson for ordinary users is straightforward: adding another person to the credential boundary increases the number of people who can expose, reuse or lose that credential.
Canada’s Privacy Commissioner Philippe Dufresne made the broader privacy principle explicit after a 2026 investigation involving ChatGPT: ‘Appropriate safeguards are the cornerstone of responsible innovation.’ That statement concerned platform privacy compliance, not password sharing, but the principle maps cleanly to account design. Separate identities are a safeguard because they let people control their own authentication, context and revocation.
The same identity logic appears in enterprise security research. Okta’s 2026 survey of 306 CISOs and senior security executives found 21% of organisations using shared credentials or broad-permission service accounts for AI-agent access. That statistic is about non-human agents, not ChatGPT users, so it should not be extrapolated as a consumer-sharing rate. Its governance lesson is still useful: shared credentials make permissions and accountability harder to scope. Okta Regional CSO Matt Immler described the desired alternative as ‘treating those AI identities as first-class identities.’
Okta CEO Todd McKinnon framed agent security around three questions: ‘Where are my agents, what can they connect to, and what can they do?’ For a personal ChatGPT account, translate that into human terms: Who can sign in? What private or connected data can the account reach? What actions can that person take? Our ChatGPT data privacy guide gives the data-layer context behind those questions.
“Appropriate safeguards are the cornerstone of responsible innovation.” — Philippe Dufresne, Privacy Commissioner of Canada, May 2026
“treating those AI identities as first-class identities.” — Matt Immler, Regional CSO, Okta, July 2026
“Where are my agents, what can they connect to, and what can they do?” — Todd McKinnon, CEO of Okta, April 2026
Four Common Claims the Documentation Does Not Support
The current search results contain several claims that sound plausible but are not supported by OpenAI’s public documentation. Separating confirmed rules from folklore is important because unsupported specifics can be more misleading than a simple ‘we do not know’.
First, there is no documented same-household exception. Some guides suggest sharing is safer or effectively permitted if both people use the same home Wi-Fi. OpenAI’s Account Sharing Policy does not make that distinction. The account is for the individual who created it. Network consistency may reduce some security friction, but it does not create a second licensed identity.
Second, there is no public fixed simultaneous-login number for one person’s devices. OpenAI says multiple devices are allowed and that usage limits may apply based on activity and subscription level. That is materially different from publishing a hard concurrency cap. An observed logout, verification prompt or temporary limit can have several causes and should not be presented as proof of a universal device threshold.
Third, a remote desktop, anti-detect browser or password-vault arrangement does not convert a personal plan into a two-user plan. Those tools may avoid directly revealing the password or may make access more convenient, but the second person is still using the account owner’s identity. A security workaround cannot rewrite the service’s account-sharing rule.
Fourth, a Business workspace does not mean everyone shares one chat history. The purpose of named seats is precisely to create managed users inside a common organisation. Workspace governance, shared resources and administrative controls can coexist with per-user identities. If your goal is to understand long-lived conversation records rather than access rights, our AI conversation archive explainer covers how histories and archives can create privacy consequences of their own.
Finally, do not confuse ‘no immediate error’ with ‘approved’. Many online account-sharing guides are written by vendors selling remote-access, anti-detect or subscription-sharing products. Their technical tips may describe observable behaviour, but their commercial incentive makes primary-source verification especially important.
A Two-Minute Decision Framework
A useful decision can be made in under two minutes by starting with identity rather than price. Ask the questions in order.
1. Are both sessions used by the same human? If yes, one account across multiple devices is consistent with OpenAI’s policy. Expect plan-dependent usage limits, and secure the account with strong authentication. If the answer is no, move to separate identities.
2. Are the two people acting as private individuals? Give each person their own Free, Go, Plus or Pro account according to their needs and local availability. Do not assume one paid personal subscription can be split into two profiles; OpenAI does not offer that model today.
3. Are they members of the same organisation and do they need shared governance? Consider Business if the minimum two-seat commitment, central billing, SSO/MFA options, workspace controls and business-data terms are useful. Compare the annual commitment against two independent personal subscriptions, not just the per-seat headline.
4. Do they only need to collaborate on content? Share the output, source files, allowed links, prompt templates or workspace resources instead of the credential. The right object to share is usually the artefact, not the identity.
5. Does either user have sensitive history, personal memory, client files or connected apps in the account? Treat that as a strong reason to keep accounts separate even if both people trust each other. Trust does not solve accidental disclosure, mixed personalisation or unclear accountability.
The final check is reversibility. Imagine the two people stop collaborating tomorrow. If removing one person requires changing a password and then wondering which chats, memories, files and connections belong to whom, the setup was not properly separated in the first place. That reversibility test also protects future collaboration: each person can keep their own history and work product while shared assets remain shareable without a messy account handover.
Our Editorial Verification Process
This article was verified on 22 September 2026 against OpenAI’s live Help Centre, Terms, consumer and Business pricing documentation, Business billing guidance, account-switching documentation and 2026 security materials. We treated OpenAI’s Account Sharing Policy as the controlling primary source for the identity-versus-device distinction and used the Terms of Use to confirm the credential-sharing restriction. Pricing figures were taken from current OpenAI pages rather than older third-party summaries.
For competitive research, we reviewed ten prominent live-ranking results for the exact query and close variants, including pages from AnyViewer, Hidemium, AIDeals, TheWearify, AccountShare, Primingo, AdsPower, Multilogin, DICloak and PandaNPC. We recorded their recurring structures and claims, then built this article independently rather than copying their section order. The largest gaps were outdated ChatGPT Team/Business pricing, unsupported numeric concurrency claims, weak separation between multiple devices and multiple humans, promotional workarounds, and little engagement with 2026 shared-LLM privacy research.
The publication’s live sitemap.xml endpoint did not return parseable sitemap content through our research environment. Following the brief’s fallback rule, we did not fabricate sitemap entries; we selected eight directly relevant Perplexity AI Magazine pages that were independently verified as live and indexed, then placed each internal link once in a different body section.
For the human-factors section, we used the peer-reviewed CHI 2026 paper by Song and colleagues, which combined a 245-person survey with 36 interviews. We report its methodology and avoid presenting its sample as globally representative. For security context, we used Okta’s 2026 CISO survey of 306 security executives and explicitly distinguish its AI-agent findings from consumer ChatGPT behaviour. We also checked that every quoted passage stayed short, attributable and directly relevant to the claim it supports.
This article was researched and drafted with AI assistance and reviewed by the Sami Ullah Khan editorial desk at Perplexity AI Magazine. All data, citations, pricing figures, and named quotes have been independently verified against primary sources before publication.
Conclusion
Two people using one personal ChatGPT login is a poor fit not because OpenAI forbids multiple devices, but because the service draws its boundary around the person who owns the account. One user can move across devices; two users should have two accounts. OpenAI does not publish a universal simultaneous-device ceiling for the account owner, so there is no sound basis for invented ‘safe’ device counts, household-IP exceptions or browser tricks.
The 2026 product landscape also makes the legitimate alternatives clearer. Two Plus accounts cost $40 per month at current public US pricing. A two-seat Business Standard workspace costs $50 on monthly billing or $40 per month equivalent on annual billing and adds organisational administration, but it is not a consumer family plan. Free and Go accounts can be enough when the real need is simply separate access.
The open question is whether OpenAI will eventually offer household or duo profiles with separate memory and history under one bill. It does not currently document such a product. Until that changes, the cleanest rule is the simplest one: share files, outputs and workspace resources when collaboration requires it, but keep human identities separate.
FAQs
Can two people use the same ChatGPT account at the same time?
Two sessions may technically remain active, but OpenAI says a personal account is for the individual who created it. The same person may use their account on multiple devices; a second person should create their own account or use a separate Business seat. OpenAI does not publish a fixed simultaneous-device limit for the account owner.
Can I use my ChatGPT account on two devices at once?
Yes, OpenAI explicitly says you may use your account on multiple devices. The published policy does not state a universal numeric concurrency ceiling. Usage limits can still depend on account activity and subscription level.
Can I share ChatGPT Plus with my spouse or family member?
OpenAI’s account-sharing policy says the account is meant for the individual who created it, so a spouse or family member should use their own account. OpenAI does not currently publish a consumer family or duo subscription that creates separate member profiles under one personal plan.
Will ChatGPT ban me for logging in from different devices?
OpenAI allows the account owner to use multiple devices, so different devices alone are not described as a violation. Security checks or usage limits can still occur. OpenAI does not publish a rule saying a particular number of devices or IP addresses automatically causes a ban.
Does ChatGPT have a family plan in 2026?
No consumer family or duo plan is listed in OpenAI’s current plan documentation. Free, Go, Plus and Pro are individual plans. Business and Enterprise are designed for organisations, with Business starting at two paid users.
Is ChatGPT Business cheaper than two Plus accounts?
At current public US prices, two Plus subscriptions total $40 per month. Two Business Standard seats total $50 on monthly billing, or $480 per year—equivalent to $40 per month—on annual billing. Business adds organisational controls and an annual commitment when billed at the lower rate.
Can one person keep two ChatGPT accounts signed in?
Yes. OpenAI’s web account switcher can keep two accounts signed in within one session, with one active at a time. Their chats, memory, files, billing and workspaces remain separate. This is designed for one user managing separate identities, such as personal and work accounts.
What is the safest way for two people to collaborate in ChatGPT?
Use separate accounts and share the artefact rather than the credential: source files, permitted conversation links, prompt templates, shared Library content or managed workspace resources. In an organisation, separate Business seats provide central administration without requiring users to share one password.
References
OpenAI. (2026). OpenAI Account Sharing Policy.
OpenAI. (2024, December 11; current terms accessed September 22, 2026). Terms of Use.
OpenAI. (2026). ChatGPT Plus.
OpenAI. (2026). ChatGPT Business: General FAQ.
OpenAI. (2026). Managing Billing and Seats in ChatGPT Business.
OpenAI. (2026). Use Multiple Accounts With Account Switching.
Song, J. E., Lee, E., Im, J., Jang, H., Kim, E., & Lee, J. (2026). “Don’t Look, But I Know You Do”: Norms and Observer Effects in Shared LLM Accounts. Proceedings of CHI ’26. ACM DOI record.
Office of the Privacy Commissioner of Canada. (2026, May 6). Joint Investigation Into OpenAI’s ChatGPT Leads to Better Protections for Canadians’ Personal Information.
Okta. (2026, July 29). Global CISO Insights 2026: Identity Security in the Age of AI.
