OpenAI's safety report: I went looking for it, it comes out in 2027

In short

• OpenAI says it made more than 200 reports to law enforcement in Q3 2026. That's the only figure in the post.
• The first public report will come out "early 2027," covering the second half of 2026, then every six months.
• The new limits (mass attacks, weapons for teens, automatic ending of conversations) will start rolling out "in the next few weeks," according to OpenAI.

The report: a promise, not a document

The post announces a report, it doesn't publish one. OpenAI commits to explaining in it how it trains its systems to detect and handle community safety issues, with anonymized examples and "relevant data". See you in early 2027.

Small detail that amuses the copy editor I am: the same text first calls it a "Community Safeguards Report," then a "Community Safety Report". Two names for a report that doesn't exist yet, that's some output ^^

The only figure, and what it measures

• What it is: more than 200 "law enforcement referrals," reports to law enforcement, cited in the sentence where OpenAI says it forwards imminent threats of violence.
• The period: Q3 2026, with nothing more specific.
• The evidence offered: OpenAI asserts that these referrals "make a positive difference" in foiling violent plots, and points to two news articles. No quantified tally.

What changes, according to OpenAI

For everyone:

• ChatGPT will give fewer details on past mass attacks: OpenAI is restricting reconstructions, details on the weapons and the equipment, and detailed comparisons of tolls that could increase an attacker's notoriety. Understanding the event and what the victims went through remains possible. That applies "regardless of the age or the reason for the request".
• If recent conversations show warning signs of violence, the responses will become more cautious about info that could be misused.
• Faced with a fixation on violence or on the idea of seriously injuring someone, ChatGPT will be able to cut the conversation off all by itself. Several cutoffs can lead to a temporary account pause.

For under-18s:

• Practical information on weapons and security systems that could be used for violence will be restricted, even with no sign of violent intent at all. OpenAI also says it is working to apply its stricter limits as soon as it estimates that an account belongs to a minor.
• Age estimation, already strengthened according to OpenAI, continually reassesses new signals. An adult misclassified can prove their age via Persona, OpenAI's verification provider.
• Parents whose account is linked to their teen's already receive an alert if the teen's account is deactivated for violence (since July). OpenAI says it is working on an alert when the account is paused for repeated discussions about violence.

OpenAI notes that teens can still use ChatGPT to prevent or de-escalate violence, seek support, or reach a trusted adult.

What the post doesn't say

I'm not asking for the moon, just the figures that would give the only one we have some meaning:

• how many spotted conversations or accounts these 200 reports fall on;
• how many accounts were banned for violence;
• how often age estimation gets it wrong, in one direction or the other;
• who will verify the future report's figures: the post mentions no outside oversight;
• a precise rollout date for each measure.

My take (this is an opinion)

The measures go in the right direction, and on a subject like this I prefer a lab that clamps down too hard to a lab that lets it slide. But calling a post a "report" when the report arrives in a few months is PR running ahead of transparency. I'll come back and read it in early 2027. With a calculator.

What we don't know yet

• The actual content of the first report, expected in early 2027.
• The exact arrival date of each measure: "in the next few weeks," nothing more.
• What the OpenAI pages linked from the post say (age, parental controls, usage rules): we have not reread them for this article.

Key terms

• Report to law enforcement: OpenAI forwards to the police a threat of violence judged imminent.
• Age prediction: a system that estimates, from signals such as usage habits, whether an account belongs to a minor.
• Under-18 experience: the version of ChatGPT with stronger protections, where a teen is moved over when the system estimates they are a minor.
• Persona: the provider that lets an adult wrongly classified as a minor prove their age.

About the source

• Who announces: OpenAI, post "Community safety report and new protections for teens," Safety category.
• Date per the source: October 30, 2026, 09:00 GMT in the official RSS feed.
• Reading: OpenAI's page refused our automated access, and the text was read in full on the Internet Archive copy from 31/10/2026.
• Links: OpenAI post · Internet Archive copy

Source passages

Show the 15 sources

Ref.	Quote (original language)	URL
S1↑	"October 30, 2026" / "Safety" / "Community safety report and new protections for teens"	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S2↑1↑2↑3↑4↑5↑6	"Our work includes bringing imminent threats of violence to the attention of law enforcement, and we made over 200 law enforcement referrals in Q3 2026."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S3↑	"These referrals are making a positive difference in disrupting violent plots" ("positive" and "difference" link to open.kg and wpbf.com)	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S4↑1↑2↑3↑4↑5	"We will publish our first report in early 2027, covering the second half of 2026, with updates every six months thereafter."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S5↑	"The report will explain how we train our systems to detect, investigate, and respond to community safety issues. We'll share anonymized examples [...] and include relevant data to support these findings."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S6↑	"a new Community Safeguards Report" / "We're committing to launch the Community Safety Report"	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S7↑1↑2	"we're tightening our policies to restrict reconstructions of past mass attacks, including details about the weapons and equipment used and how they were obtained, and detailed comparisons on scale of harm that could amplify an attacker's notoriety. These limits apply regardless of a user's age or reason for asking."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S8↑1↑2	"when recent chats show warning signs of possible violence, ChatGPT will respond more conservatively" / "We're also working to apply those stricter limits when we predict someone is under 18, even if their chats show no signs of violent intent."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S9↑	"ChatGPT will be able to automatically end the chat to interrupt that pattern in the moment. If ChatGPT ends multiple chats for that reason, we may temporarily pause the account."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S10↑	"We're extending those protections to restrict practical information about weapons and security systems that could be misused to facilitate violence, even when there's no sign of violent intent."	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S11↑1↑2↑3↑4	"analyzes multiple signals such as behavior patterns" / "We strengthened our age prediction system to continually assess new signals" / "move teens into the under 18 experience when appropriate" / "through Persona, our age verification partner"	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S12↑	"In July, we expanded those alerts to notify parents if their teen's account has been deactivated for violating our usage policies on violence. We are working to introduce notifications for parents if we pause their teen's account"	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S13↑1↑2↑3↑4	"The safeguards detailed here will begin rolling out in the next few weeks"	https://openai.com/index/community-safety-report-and-new-protections-for-teens
S14↑	"<pubDate>Fri, 30 Oct 2026 09:00:00 GMT</pubDate>"	https://openai.com/news/rss.xml
S15↑	"Teens can still use ChatGPT to help prevent and de-escalate violence, seek support, and connect with a trusted adult."	https://openai.com/index/community-safety-report-and-new-protections-for-teens