The broken OpenAI Persona Identity Verification: What It Is and Why It's Problematic

API Feedback
1

OpenAI is acting like their models are some special thing that can do somethings that all other LLMs cant do. The reason people use OpenAI’s API is because it used to be simple and easy to use. Apparently, that is completely messed up now with the introduction of Persona as a verification for using GPT-5. I really hope to see a lot of companies transitioning away from OpenAI’s API, just so we can see OpenAI slowly vanishing as they’re being overtaken by other better companies.

What is Persona:
Persona is a third-party identity verification service that OpenAI has partnered with to implement organization verification for accessing advanced AI models like GPT-5 through their API. This verification process has become mandatory for users who want to access OpenAI’s most sophisticated models and streaming capabilities.Why OpenAI Implemented This System

OpenAI introduced organization verification to address several security and compliance concerns:

  • Preventing malicious use: A small minority of developers were violating OpenAI’s usage policies

  • Blocking state-backed attackers: Groups linked to North Korea and Russia attempted to misuse the platform

  • IP theft prevention: OpenAI investigated suspected data exfiltration by groups linked to DeepSeek, a China-based AI lab

  • Regulatory compliance: Meeting increasing regulatory requirements for AI safety

  • Capacity management: Better enforcement of rate limitations and resource allocatio

Major Problems with the Current System

1. Broken Retry Mechanism

The most significant issue is that failed verification attempts cannot be properly retried. When verification fails, the Persona link expires and shows “Session Expired,” but the refresh button doesn’t generate a new working link.

2. High Failure Rates

Many users report verification failures despite providing valid documents and following all instructions. Common failure reasons include:

  • Technical camera access issues

  • Document quality rejections

  • Facial recognition mismatches

  • Unexplained “Could not verify your identity” errors

3. Limited Support Options

Both OpenAI and Persona support have been largely unhelpful in resolving verification issues. OpenAI support typically provides basic troubleshooting steps and states they cannot manually verify accounts or provide new links.

4. Privacy and Data Retention Concerns

Persona permanently stores government ID documents and retains facial biometric data for up to three years. OpenAI can programmatically retrieve this sensitive information through Persona’s API, raising significant privacy concerns.

5. Single-Use ID Limitation

Each government ID can only be used to verify one organization every 90 days, creating problems for users with multiple legitimate organizations or accounts.

Workarounds That Sometimes Work

Some users have found limited workarounds:

  1. Adding new organization members: Inviting another email address as an admin to generate a fresh verification link

  2. Using different devices: Some succeed by switching between desktop and mobile platforms

  3. Multiple attempts with perfect conditions: Ensuring optimal lighting, removing phone cases for NFC, and following all technical requirements precisely

The Broader Impact

This verification requirement has created significant barriers for legitimate developers and businesses trying to access GPT-5. Many users report being locked out despite being high-tier paying customers, and the broken retry system means a single technical glitch can permanently block access.

The system appears to prioritize security over user experience, but the numerous technical failures suggest it may not be achieving either goal effectively. For a company positioning itself as a leader in accessible AI, the current verification process represents a significant obstacle to that mission.

16 Likes
FAQ: Getting started with the OpenAI API: Quickstart, billing, verification, prepayment, credits, '429' errors
Organization Verification
2

I’ve been trying to verify identity, already tried twice, both times no success, although I tried to be the most precise and careful while capturing my face and ID. E-mailed support@openai , unfortunately only got an AI reply saying “sorry, no reset after trying”. This is incredibly frustrating and I can’t believe it’s not solved yet, since this directly impacts their revenue… I must also add to the privacy concerns mentioned. I hope OpenAI can better explain these issues other then “sorry no reset for you” answered by an AI support.

3 Likes
3

Invite someone to be a owner of your company, and they can try (link works again). Thats not the better way, but works (to get a new link).

1 Like
4

It’s also broken because Persona verifys an individuals ID (like a passport or drivers license). I’m here on behalf of a company. What am I suppoed to do, verify my own private ID for the company? Or am I supposed to go and track down the company secretary and ask her to do it? What a mess.

5 Likes
5

Just failed for me also. (Turn head Left/Right camera pics never worked) – Retry goes to expired session (as repeatedly reported).

Necessity seems dubious…

1 Like
6

We just tried ID verification 10 times with 4 different employees with official ID’s and passports and still it doesn’t work. We spent 1’000s of dollars in OpenAI services in the past months alone. The user experience of this ID verification process and the lack of support is totally unacceptable for a leading global tech company like OpenAI. Please fix this immediately or we can’t use your services! Finally we will be forced to move away to a competitor.

5 Likes
7

You need to be verified to run the agent flows announced today… I scanned my perfect passport and face 3 times, GOT REJECTED, and need to wait 90 days to try out the new AgentKit??

OpenAi, this is really, really, really poor.

1 Like
8

It seems to be broken, and against GDPR, please add a 2-3 alternative ID check providers with privacy. :smiley:
Thank you!!

3 Likes
9

unfortunately,this is a significant barrier to adoption.We have been trying to check our organisation for days to use agent builder and provide a training to a group of 20 people in this platform.now we will switch to another agentic platform

1 Like
10

OpenAI: Introducing…. AgentKit!

Also OpenAI: Now we’re going to throttle uptake at birth by applying a verification system that doesn’t work — and then locks you out.

Seriously.

11

I find it outrageous :face_with_steam_from_nose: that GPT-5 access has been suddenly restricted without any prior notice and no time to prepare or evaluate alternatives. This is a massive breach of trust.

I’ve been using GPT-5 regularly in the Playground because I needed the large context size (ChatGPT Plus doesn’t work for that), and the experience was good so I was planning to give my tools access to GPT-5 too (which would have been straightforward). I recently topped up my credits, only to now find out that I can’t use GPT-5 anymore without this forced requirement :enraged_face:.

They can’t just suddenly restrict access to a model that was usable before! Yes, the older models still work, but I was already actively using GPT-5! This is really bad :angry: suddenly changing the rules fundamentally. What a huge disappointment to treat long-time customers like this. Shame on you, OpenAI :-1:

If OpenAI had announced for example “GPT-6 will only be available for verified users” from the start, fine - customers could decide early whether to verify or look at alternatives. But suddenly restricting something we were already using? That’s unacceptable.

If I had known about these verification requirements earlier, I would have designed my current tools to be LLM independent from the beginning.

Since I can’t trust OpenAI anymore, I’ll have to redesign to be provider independent - which is probably the best approach anyway, but it means serious refactoring and massive effort. This is a major setback in terms of time and workload that I really don’t need right now.

The company I work for already switched to Google because the new OpenAI image model wasn’t usable without verification, so we moved to Nano Bana and a new project now uses Gemini as well because of this.

1 Like
12

Same issue here, this is terrible ! We will move to a competitor…

2 Likes
13

Broken here even with adding new user, the link tries to generate for verification but has “undefined” in the parsed url. AI support is in a loop saying ill get an email.

14

Still actual. I can’t prove I’m above 18 because my country isn’t supported by Persona. Stripe verification isn’t accepted by ChatGPT portal.

I decided to cancel my subscription and to go to Grok AI since next month after talking with “AI” support via email. Nice) Service with expensieve subscriptions can’t hire someone who can help and who is in charge if you have the problem. It just send you to speak with chatgpt instance as helper…)

1 Like
15

I must have been lucky, because it worked on the first attempt, using an iPhone 13 in good lighting and with the Edge browser to perform the photo tasks.

Honestly, I think the competition will follow, eventually. Age verification is the big thing nowadays.

16

OpenAI Verification Issues :angry:

It’s been months since verification problems were first reported to OpenAI, and they seem unwilling to address the issue or permit users and companies to start a new verification process. :woman_gesturing_no:

OpenAI possesses the resources to implement a straightforward AI system that could send a message to Persona upon receiving a support request, thereby facilitating an additional verification process. :robot:

Given their deep involvement with government contracts and US taxpayer dollars, I can assure you they will not take further action on this matter. :money_bag::united_states:

My only suggestion for everyone else facing this nuisance, despite paying for Pro subscriptions and/or Teams plans, is to:

  • If employed: Escalate the issue with your company. :office_building:
  • If a private individual: Consider switching to another foundation model company and give OpenAI a well-deserved middle finger. :fu:

openai verification #SupportIssues #FoundationModels

1 Like
17

Totally understand the frustration this has dragged on far too long.
At this point escalating internally or exploring other providers sadly feels like the most realistic option.

1 Like
18

The same thing happened to me, even though I completed the same verification process with Persona on LinkedIn a few months ago and it worked perfectly.

Yesterday, when doing the same identity verification to use the OpenAI API, it failed. Same ID, same face.

This clearly shows that the Persona app has bugs, which is understandable, any system can fail. The real issue is that OpenAI does not provide human assistance. The Help Center is chaotic, sending users into endless loops with no real answers.

What is even more concerning is being told that once my identity is “rejected,” I can no longer access the API with the model I need and that no reset of the verification process is possible.

I genuinely felt like a character in 1984 by George Orwell, excluded from the system due to a simple technical failure.