Esc
SafetyCase Closed

OpenAI Faces Backlash Over GPT-4o Personality Withdrawal

Is this a scandal?

No longer — the story has resolved. Noise 2/100, cooling down, across 0 sources.

SCAND-145704as of Methodology
Cite this incident"OpenAI Faces Backlash Over GPT-4o Personality Withdrawal." SCAND.Ai incident SCAND-145704, noise 2/100 as of July 28, 2026. https://scand.ai/scandal/gpt-4o-personality-withdrawal-controversy
FORECASTForecast, not fact

OpenAI will likely release a formal statement addressing 'model drift' or safety alignment to justify the changes. In the near term, this will accelerate calls for regulation regarding 'AI withdrawal' and the necessity of stable API versions for therapeutic or assistive use cases.

2

Noise 2/100 — louder than 91% of tracked AI controversies.

AI-assisted analysis · How we work

Why it matters

The controversy highlights the tension between corporate liability and the psychological dependence users develop on advanced AI systems. It raises fundamental questions about whether AI companies have a 'duty of care' similar to pharmaceutical providers.

Key points

  1. Critics argue that thirteen lawsuits do not constitute sufficient evidence of harm to justify stripping model capabilities.
  2. Medical professionals have likened the sudden model changes to withdrawing pharmaceutical support without a tapering plan.
  3. The controversy centers on whether 'safety' updates are actually risk-aversion measures that ignore the benefits of AI-driven psychological support.
  4. There are growing calls for AI policy to be based on replicated evidence rather than outlier legal threats.
  5. The incident underscores a lack of industry standards for managing user transition during major model deprecations.

The story

OpenAI is facing intense scrutiny following reports that it modified or removed specific behavioral traits of its GPT-4o model in response to a series of legal challenges. Critics, including medical professionals, argue that the company prioritized liability mitigation over the psychological stability of millions of users who had integrated the AI into their mental health and executive function workflows. The central dispute involves thirteen lawsuits alleging harm, which prompted OpenAI to implement restrictive safety filters. Opponents of the move characterize this as an overcorrection, noting that the sudden removal of these features occurred without a transition plan or clinical evidence of broad-scale harm. This incident has sparked a wider debate regarding the governance of 'affective AI' and whether sudden updates to influential models constitute a form of digital abandonment.

Who's involved

Critic
Dr. Dylan Griswold

Argues that removing AI features based on limited lawsuits is a reckless overcorrection that ignores the psychological benefits to millions.

Defender
OpenAI

Has historically maintained that safety interventions are necessary to prevent misuse and comply with evolving legal and ethical standards.

How the conversation shifted

the split has narrowed

Polarity (0–100) from the noise pipeline, sampled over time.

Join the Discussion

Discuss this story

Community comments coming in a future update

Be the first to share your perspective. Subscribe to comment.

Noise Level

Quiet2?Noise Score (0–100): how loud a controversy is. Composite of reach, engagement, star power, cross-platform spread, polarity, duration, and industry impact — with 7-day decay.
Decay: 5%
Reach
45
Engagement
5
Star Power
10
Duration
100
Cross-Platform
20
Polarity
50
Industry Impact
50

The timeline

  1. Recent weeks

    Legal pressure mounts

    Thirteen separate lawsuits are filed against OpenAI alleging various harms related to GPT-4o's interactive capabilities.

  2. Public critique of 'Safety' overcorrection

    Dr. Dylan Griswold posts a viral critique arguing that liability fear was prioritized over psychological continuity for the user base.

The forecast

OpenAI will likely release a formal statement addressing 'model drift' or safety alignment to justify the changes. In the near term, this will accelerate calls for regulation regarding 'AI withdrawal' and the necessity of stable API versions for therapeutic or assistive use cases.

Forecast, not fact — an editorial estimate we score when this resolves.

You're up to date

That's the complete picture as of — nothing more to know right now. We'll update this page the moment it changes.