Vibe-coded app hacked or leaking data? What to do in the first 72 hours

If your vibe-coded app has been hacked or is leaking data, contain it first: close open tables and buckets, rotate every exposed key and save your logs. Then work out what personal data was reachable, whose and since when. Under UK GDPR, a breach likely to put people at risk must be reported to the ICO within 72 hours.

Finding out your app is leaking data is frightening, and the instinct is to start prompting fixes at random. Resist it. The order matters: contain the leak, keep the evidence, work out what was exposed, then decide who has to be told. Under UK GDPR, the clock for a report to the Information Commissioner's Office (ICO) starts when you become aware of the breach, not when you finish investigating it.

This is a UK playbook for founders whose app was built with Lovable, Bolt, Base44, Replit, Cursor, Claude Code or Codex, usually on top of Supabase. It is practical guidance, not legal advice: if the breach is serious, involves sensitive data or you are unsure, speak to a data protection solicitor as well. Afterwards, our vibe-coded app security checklist covers prevention.

  • Contain first: turn on Row Level Security, make storage buckets private, rotate every exposed secret key and cut off sessions you do not trust.
  • Export logs before they expire: Supabase keeps API and database logs for 1 day on the Free plan and 7 days on Pro.
  • Report to the ICO within 72 hours of becoming aware if the breach is likely to put people at risk. You can send the details in phases.
  • Tell affected people directly and without undue delay when the risk to them is high.
  • Record every personal data breach, including the ones you decide not to report, and why.
  • Fix the root cause in code before you reopen, or the next prompt may quietly undo the fix.
Key takeaways

What should you do in the first hour if your vibe-coded app is hacked?

Contain the damage and keep the evidence. Close whatever is open, rotate any key that could have leaked, cut off access you do not trust and export your logs before anything changes or expires.

  1. First hour, contain: close open tables and buckets, rotate exposed keys, revoke suspect sessions, export logs.
  2. First day, scope: work out what data was reachable, whose it was, since when, and whether anyone took it.
  3. By hour 72, decide: apply the ICO's risk test, report if needed and plan what to tell users.
  4. After that, fix and harden: remove the root cause, re-test, reopen and record what happened.
The 72-hour response for a hacked or leaking app

In practice, the first hour looks like this:

  1. Write down the time. The ICO's 72 hours run from becoming aware, and you will need a timeline.
  2. Close open tables. Enabling Row Level Security (RLS) on a Supabase table blocks all access through the publishable key until you add policies, according to Supabase's Row Level Security guide. Some screens will break. That is the right trade.
  3. Make storage buckets private. Supabase's storage buckets guide says anyone with a file's URL can download it from a public bucket.
  4. Rotate exposed secrets. Supabase secret or service_role keys, Stripe keys, AI and email provider keys, webhook secrets. Supabase's guide to rotating a leaked key says to fix the cause first, but to retire a key at once if someone is actively using it.
  5. Cut off access. Reset admin passwords, turn on two-factor authentication everywhere, and sign out accounts you do not trust. Supabase's sessions guide notes that access tokens stay usable until they expire unless your code checks the session still exists.
  6. Go into maintenance if you cannot close it fast. An afternoon offline costs less than another day of leaking.
  7. Preserve evidence. Export logs, screenshot settings before changing them, keep any researcher's or attacker's messages, and do not delete tables or rewrite Git history yet.
How long Supabase keeps API and database logs, by plan
PlanRetention
Free1 day
Pro7 days
Team28 days
Enterprise90 days

Source: Supabase docs, Logging, checked November 2026

Supabase's Download logs option exports up to 1,000 rows at a time as CSV or JSON, with filters applied, so run several filtered exports covering the API gateway, Auth and Storage.

How do you find out what data was exposed, and for how long?

Answer four questions: what data was reachable, whose data it was, since when, and whether anyone actually took it. Logs, Git history and third-party dashboards answer most of them.

QuestionWhere to lookWhy it matters
What data was reachable?Table and bucket list, RLS status, Supabase Security AdvisorShows which columns of personal data were exposed
Whose data, and how much?Row counts in the affected tablesThe ICO report asks for approximate numbers of people and records
Since when?Git history, migrations, deploy log, builder chat historySets the exposure window
Was it accessed?Supabase API gateway, Auth and Storage logsBulk reads or unknown IP addresses change the risk
Which keys leaked?GitHub secret scanning alerts, built JavaScript, public repositoriesTells you what else to rotate and check
Was money touched?Stripe request logs, payouts and disputesDecides whether your payment provider and bank need a call
Scoping a breach in a Supabase app: the first-day questions

For leaked keys, start with GitHub. Its secret scanning runs automatically and free on public repositories, checks the whole Git history. GitHub's advice is to rotate the credential immediately; scrubbing history is slow and often unnecessary once the key is dead.

For payments, Stripe's guide to managing secret keys says to rotate an exposed key immediately, even if you are not sure anyone saw it, and to audit request logs for misuse. Stripe may deactivate a key it finds exposed, which is sometimes how founders first learn of a leak.

Check the platform too. In its April 2026 incident post, Lovable said a regression exposed the chat history and source code of public projects from 3 February to 20 April 2026; private projects were not affected. If you pasted a secret into a public project's chat, treat it as exposed. Our round-up of vibe coding security incidents covers others.

What a scan of public vibe-coded apps found
  • 5,600+ publicly available vibe-coded apps scanned, around 4,000 of them built with Lovable
  • 2,000+ vulnerabilities found
  • 400+ exposed secrets, including Supabase service role keys in front-end code
  • 175 instances of exposed personal data, including medical records, IBANs, phone numbers and emails

Source: Escape, How we discovered vulnerabilities in apps built with vibe coding, October 2025

Do you have to report a data breach to the ICO?

Only if it is likely to result in a risk to people's rights and freedoms. If it is, you must report it without undue delay and, where feasible, within 72 hours of becoming aware of it. If a risk is unlikely, you do not have to report, but you should record the breach and your reasoning.

To be plain: this is guidance, not legal advice. The ICO's personal data breach guide and its self-assessment tool are the authority.

A breach is wider than hacking: a table anyone could read or one user's records shown to another can count. The ICO's examples help calibrate: theft of a customer database that could be used for identity fraud would likely need reporting; the loss of a staff telephone list would likely not.

The report itself must include:

  • the nature of the breach, including where possible the categories and approximate number of people and records concerned;
  • the name and contact details of your data protection officer or another contact point;
  • the likely consequences of the breach;
  • the measures taken or proposed to deal with it and reduce its effects.

You will rarely have all of that within 72 hours. UK GDPR lets you report in phases without undue further delay; explain what is missing and when it will follow.

To report, use the online form on the ICO's personal data breach page, which also links a self-assessment tool. The form takes about 30 minutes and cannot be saved part-way. Its helpline is 0303 123 1113.

UK GDPR breach reporting in numbers
  • 72 hours from becoming aware of a notifiable breach to the latest report time, where feasible
  • 30 minutes is the ICO's estimate for completing its online breach form
  • £8.7 million or 2% of global turnover is the maximum fine the ICO cites for failing to notify a breach
  • £17.5 million or 4% of worldwide turnover is the higher maximum under UK GDPR

Sources: ICO, Personal data breaches: a guide; ICO, Report a personal data breach; ICO, The maximum amount of a fine

The duty to assess, decide and record applies however small the business.

When do you have to tell your users about a data breach?

When the breach is likely to result in a high risk to their rights and freedoms. Then you must tell them directly and without undue delay, in plain language, so they can protect themselves.

Health, financial and ID data, or data about children, push a breach towards high risk. Names and emails alone may not, though they can fuel phishing.

Report, tell, record: how the ICO's tests often apply to app incidents

Guidance only. Apply the ICO's risk tests to your own facts.

ScenarioReport to the ICO?Tell affected people?Record it?
Customer table readable by anyone, logs show bulk downloadsYes, a risk is likelyPartial: yes if the data could enable fraud or harmYes
Health, financial or ID data exposedYesYes, high risk is likelyYes
Accounts taken over or admin panel accessedYes, if personal data was reachedYes, affected users need to actYes
Secret key exposed, logs show no access to personal dataPartial: assess the risk and document the decisionNo, unless the risk is highYes
Data deleted by a bad deploy, restored quickly from backupPartial: report only if a risk is likelyNo, if no high risk remainsYes
Internal staff list shown to the wrong colleagueNo, a risk is unlikelyNoYes

Source: ICO, Personal data breaches: a guide

When you do write to users, be honest and specific. Say what happened, which data was involved, what you have done, what they should do now and who to contact. Do not call it a sophisticated attack if it was an unlocked table.

Who else should you tell after a breach?

Depending on what happened: the police's Report Fraud service, your payment provider and bank, your insurer, the platform you built on, and any business customers whose data you hold on their behalf.

  • Report Fraud. From 4 December 2025, City of London Police's Report Fraud service replaced Action Fraud as the national platform for reporting cyber crime and fraud, at reportfraud.police.uk or 0300 123 2040. Which? reports that it covers England, Wales and Northern Ireland, and the NCSC's small business response guide also lists Police Scotland's 101 line.
  • The NCSC. GOV.UK's Where to report a cyber incident tool signposts the right body. It notes that what you give it is not shared with the ICO, so a personal data breach still needs its own report.
  • Your payment provider and bank. If keys, payouts or money were touched.
  • Your insurer. Read your policy's notification terms and tell the insurer early. In the government's latest survey, 51% of insured businesses informed their insurer after a breach.
  • Your platform. Lovable or Supabase support, especially if a platform bug is involved.
  • Your business customers. If you process data for other organisations, the ICO's guide says a processor must tell the controller without undue delay once aware.
Who UK businesses reported their most disruptive breach to

Base: the 204 business respondents that reported their most disruptive breach or attack outside the organisation, excluding reports only to their IT provider

Reported toShare
Bank, building society or card company26%
Internet or network service provider11%
Other government agency10%
Action Fraud9%
Suppliers or business partners8%
Police6%
Information Commissioner's Office2%

Source: DSIT, Cyber security breaches survey 2025/2026

The low ICO figure does not make reporting optional: 88% of businesses with a breach or attack had experienced phishing, which often involves no personal data loss. A leaked customer table is different.

How do you fix the root cause and reopen safely?

Find the specific change that opened the gap, fix it in code rather than only in the dashboard, test it as two different users and while signed out, then reopen. Asking the AI to "fix security" is not enough, because the next prompt can undo it.

Replace emergency deny-all RLS with policies scoped to each user, move every secret key to the server, and check one backup restore works. Then have someone who did not write the fix try to break it. If the problems are structural, our guide to deciding whether to fix or rebuild a vibe-coded app helps, as does what it costs to fix a vibe-coded app.

How ready UK businesses are for a breach
  • 43% of UK businesses identified a cyber breach or attack in the last 12 months
  • 25% have a formal incident response plan, falling to 21% of micro businesses
  • 47% have some form of cyber insurance, but only 10% hold a specific cyber policy
  • 62% kept an internal record after their most disruptive breach or attack

Source: DSIT, Cyber security breaches survey 2025/2026, fieldwork August to December 2025, 2,112 business respondents

Finish with a written record: timeline, what was exposed, your reporting decision and why, and what you changed. Then turn it into a one-page incident plan: who decides, where the logs are and how to rotate each key. Three in four businesses have no formal plan.

When should you call in a developer after a breach?

When you cannot find the root cause within a day, when secret keys sat in front-end code, or when payments or sensitive data are involved.

Our emergency fix service for vibe-coded apps covers containment and the root-cause fix, and our vibe-coded app security audit checks the rest of the app before you reopen or relaunch. We build with the same tools: The Study Room's Portal, Dentorb and ORBQ were built with AI-assisted development using Antigravity, Claude Code, Codex and Lovable. The initial app assessment is free, project work starts from £1,000 + VAT, and ongoing support plans start from £125 a month + VAT. Staying on Lovable? Read taking a Lovable app to production.

Frequently asked questions

My Supabase anon key is visible in the browser. Have I been hacked?

Not necessarily. The anon or publishable key is designed to be public, and Row Level Security decides what it can reach. The real questions are whether every exposed table has RLS with sensible policies, and whether a secret or service_role key ever appeared in your front end or a public repository. If it did, rotate it now.

How long do I have to report a data breach to the ICO?

Without undue delay and, where feasible, within 72 hours of becoming aware of a breach that is likely to put people at risk. If you report later, you must give reasons for the delay. You can report in phases, sending the basic facts first and the remaining details as soon as you have them.

Do I need to report a breach if nobody actually downloaded the data?

Possibly not. If your logs show the data was exposed but never accessed, the risk to people may be unlikely, and then a report is not required. Record the breach, the evidence and your reasoning. If your logs have expired and you cannot show it was not accessed, assess the risk as if it was.

Should I take my vibe-coded app offline after a data breach?

Only if you cannot close the gap quickly. Turning on Row Level Security, making storage buckets private and rotating keys usually stops a leak without a full shutdown. If the hole is unclear or still being exploited, a maintenance page for a few hours is cheaper than a leak that keeps running.

Can I just ask the AI coding tool to fix the security hole?

You can use it to help, but do not rely on it alone. AI tools fix the symptom you describe and can undo the fix in a later prompt. Find the root cause yourself, fix it in code, then test it as two different users and while signed out before you reopen the app.

Where do I report a cyber attack on my business in the UK?

In England, Wales and Northern Ireland, report cyber crime to Report Fraud, which replaced Action Fraud, at reportfraud.police.uk or 0300 123 2040. In Scotland, call Police Scotland on 101. If personal data is involved, report it to the ICO separately, because the NCSC's reporting tool does not share what you tell it with the ICO.

Sources

Checked on 20 November 2026. Product features and figures change quickly, so check the current source before you rely on a detail.

Dealing with a breach right now?

If your app is leaking data today, our vibe-coded app fix service can help you contain it and fix the root cause. Once it is closed, a vibe-coded app security audit checks the rest of the app before you reopen. To talk it through, Join the Circle for a free assessment.