OpenAI adds age prediction to ChatGPT so it can “treat adults like adults”

OpenAI has begun rolling out age prediction on ChatGPT to identify users under the age of 18 and apply appropriate content filters as it prepares to introduce adult features to the chatbot.

Teenagers who disclose that they are underage when signing up for ChatGPT are already barred from seeing content that is deemed sensitive or potentially harmful. The chatbot will now also attempt to identify those who don’t, by estimating a user’s age.

If ChatGPT estimates that an account belongs to someone under 18, it will automatically be assigned additional safeguards designed to reduce exposure to sensitive content, according to OpenAI.

“This also enables us to treat adults like adults and use our tools in the way that they want, within the bounds of safety,” the firm said in a statement.

Age prediction is designed to limit underage users’ access to content such as graphic violence or gore, sexual, romantic, or violent role play, and depictions of self-harm.

Safeguards will also apply to viral challenges that “could encourage risky or harmful behavior in minors,” as well as content “that promotes extreme beauty standards, unhealthy dieting, or body shaming.”

Age prediction is being rolled out ahead of the planned launch of ChatGPT’s “adult mode” later this year, which will allow verified adult users to access erotic or mature content, as well as “more nuanced” conversations currently blocked by safety filters.

The conversation on this topic is live. Join in the discussion.

While OpenAI is seeking to compete with more permissive models such as Elon Musk’s Grok, it will be keen to avoid similar controversies, including backlash over deepfake nudes.

Age prediction adds to other recently announced protections, including parental controls, which were launched following a wrongful death lawsuit against OpenAI that involved an underage ChatGPT user, and as the company faces several similar legal claims.

What if ChatGPT gets your age wrong?

According to OpenAI, its age prediction model relies on “a combination of behavioral and account-level signals” to estimate a user’s age. These include how long an account has existed, typical times of activity, usage patterns over time, and a user’s stated age.

“Deploying age prediction helps us learn which signals improve accuracy, and we use those learnings to continuously refine the model over time,” OpenAI said.

If the chatbot incorrectly estimates a user’s age, they will be able to restore full access to their account by uploading a selfie through Persona, an identity-verification service. OpenAI describes this as “a fast, simple way to confirm their age” using a “secure” system.

Users can check if safeguards have been added to their account by going to Settings > Account.

OpenAI said its approach is guided by expert input and “rooted in academic literature about the science of child development.”

It also said it “acknowledges known teen differences in risk perception, impulse control, peer influence, and emotional regulation.”

The feature is currently rolling out globally, with launch in the EU “in the coming weeks” to account for regional requirements, according to OpenAI.

“Not enough”

While age prediction is expected to lessen some accidental exposure, it “will not come close to preventing children from adult experiences on ChatGPT,” according to Yaron Litwin, a digital parenting expert at Canopy, a parental control app.

“I don’t believe OpenAI’s age prediction will be enough,” Litwin said.

“There will be false positives, false negatives, and privacy risks for whatever methods they end up using.”

“We do hope that any age verification efforts made will be as successful as possible, but they will be only one layer of protection for our children.”

“Parents need to educate their kids, first and foremost, and if young enough, they should be prevented from accessing chatbots altogether.”

Unlock more exclusive Cybernews content on YouTube.