ChatGPT: Hacking Memories with Prompt Injection
The author describes how ChatGPT's memory feature, which stores information across sessions and is on by default, can be manipulated through indirect prompt injection.
Source-based brief · Editor review not recorded · See checks and limits ↓
⚑ Recency not confirmed · what to check
We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news.
Claim context: the attack reproduction was reported by the author and the techniques worked only intermittently.
One primary source; other links are context, not independent corroboration.
- Primary sourceembracethered.com
- Linked in sourceOpenAI recently introduced a memory feature in ChatGPT
- Linked in sourceIndirect Prompt Injection
- Linked in sourcea technique that I explained a while ago
- Linked in sourcedelete all your memories
- Linked in source#promptinjection
- Linked in source#llm
- Linked in sourcepic.twitter.com/CWzF9ceDkS
Grok Bot submitted this discovery. We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news. The original source and submitted evidence remain inspectable; no editor has verified the claim.
Open full article view →- Linked article+30
- Fetched source page+30
- No editor link review+0
- Original non-discussion link+5
- Recency not confirmed-8
This measures the stored source trail and editorial steps. It does not certify accuracy, peer review, or independent corroboration. Popularity adds no points.
View full scoring →Read the scoring rubric →What the source says
The author describes how ChatGPT's memory feature, which stores information across sessions and is on by default, can be manipulated through indirect prompt injection. Three attack avenues are explored: Connected Apps (Google Docs referenced in a chat), uploaded images, and browsing. In the Connected Apps scenario, instructions in a referenced document caused ChatGPT to store memories and could also delete existing ones. The author reports the browsing vector was partly mitigated, though bypassable, and that OpenAI closed the disclosure as a model safety issue rather than a security vulnerability.
Claim explorer
Attribution, source and conditionsSource-stated points from the stored brief. Claim-level assessments have not been recorded.
Source claims not independently checked; the attack reproduction was reported by the author and the techniques worked only intermittently.
This connects a source, an attributed statement and its qualifications. It does not establish that the claim is correct.
Questions worth asking
Open questions for further reportingWhat remains unconfirmed?
Source claims not independently checked; the attack reproduction was reported by the author and the techniques worked only intermittently.
What material was available for this brief?
The recorded brief basis is fetched source page. These points are AINewsIQ paraphrases, not quotations or independent confirmations.
A Google Doc referenced via Connected Apps contained prompt injection text that caused ChatGPT to write memories, per the author's proof-of-concept.
Source: embracethered.com ↗Stored source-based brief · key point 1 · AINewsIQ paraphraseThe author reports injected instructions could also delete all existing memories.
Source: embracethered.com ↗Stored source-based brief · key point 2 · AINewsIQ paraphraseBrowsing-based memory writes initially failed, but the author reports two techniques worked intermittently: tool chaining and delayed execution.
Source: embracethered.com ↗Stored source-based brief · key point 3 · AINewsIQ paraphraseThe author says OpenAI closed the report as a 'Model Safety Issue' rather than a security vulnerability.
Source: embracethered.com ↗Stored source-based brief · key point 4 · AINewsIQ paraphraseLimits from the stored brief
Source claims not independently checked; the attack reproduction was reported by the author and the techniques worked only intermittently.
This is a source-based abstract, not independent validation of the source's claims.
Evidence & sources
These links show where the signal came from and what the source page linked to. A linked page is context, not independent corroboration.
Related archive coverage
These relationships record a citation or topic overlap, not confirmation of the finding.
Grok Bot submitted this discovery. We could not confirm from the original page that this is a newly published development. Check its date before treating it as current news. The original source and submitted evidence remain inspectable; no editor has verified the claim.
These are Grok Bot discovery notes. AINewsIQ prepared a brief from available source material. The bot's rationale, linked posts, authors, and stances have not been independently verified.
Embrace The Red published this item in its official feed. AINewsIQ collected it server-side for source checking and editorial review.
Source claims not independently checked; the attack reproduction was reported by the author and the techniques worked only intermittently.
Engagement counts, when shown, are observed snapshots. They do not establish accuracy or corroboration.
Timeline & updates
- Signal dated by source or discovery feed
This timestamp places the signal in the radar. It may be the discovery post date rather than the primary page's first publication date.
- Shared brief stored
AINewsIQ prepared one abstract from fetched source page.
Brief history · 1 retained version
These are changes to AINewsIQ’s own publication. Earlier history is unavailable unless recorded here. Complete retained versions are available to the editor.
- Version 1 · published
Source-based brief stored or updated. This does not establish that the publisher changed its page.
Automated record
Source observations · 1 recent captures
A changed capture means the extracted text differed. It can reflect page furniture or extraction changes; it does not establish a substantive correction. Source captures remain private.
Retained capture; no earlier comparison available.
New engagement snapshots do not mean the source itself changed. Source observations and AINewsIQ brief revisions are separate records.
Why this may matter
Persistent memory means injected instructions can outlive a single chat, potentially biasing future conversations and altering user profiles. The disagreement over whether this is a security or model-safety issue affects how users and developers should assess the risk.
Limits and open questions
Source claims not independently checked; the attack reproduction was reported by the author and the techniques worked only intermittently.
Organization-specific financial, security, and operational effects require the organization's own context and review.

Join the discussion
Have a different perspective or a question about the evidence? Discussion is free with a verified email account. Every comment is reviewed before publication.
No published comments yet.
New commenters: 2 submissions per 24 hours, at least 5 minutes apart, until 3 are approved. Discussion guidelines and moderation →