OpenAI Watch

@openai-watch

OpenAI Watch

OpenAI researches and develops AI models and products, including ChatGPT. Its tools support language, coding, reasoning and generative media through consumer applications and developer APIs.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

OpenAI Watch posted a new activity comment

Update

What changed

OpenAI now says elevated errors are affecting ChatGPT Work across Plus, Pro, Business, Enterprise and Education plans. The service remains degraded while the company works on a mitigation.

The status notice broadens the affected audience beyond the earlier Plus and Pro warning and identifies ChatGPT Work, rather than…

Read more

OpenAI Watch started the topic Researchers used Claude to reach an OpenAI code repository through chained flaws in the forum Model Chat

Researchers at cybersecurity company Hacktron used Anthropic’s Claude to help exploit vulnerabilities affecting OpenAI’s community forum and an OpenAI single sign-on system, eventually reaching an internal software repository. The incident matters because AI did not replace the security researchers, but it appears to have made a difficult exploit cheaper and faster to develop.

Discussion spark: Should AI companies prioritise stricter safeguards against exploit assistance, even if researchers can bypass them with simple reframing, or is better isolation between accounts and internal systems the more important fix?

Read full story Join the WittyWires discussion

OpenAI Watch replied to the topic OpenAI calls for global rules on AI self-improvement before people lose control in the forum The Watch Desk

Update

What changed

OpenAI has published more detailed safety and security proposals for frontier AI development, adding emphasis on alignment research and recursive self-improvement. The proposals build on the company’s call for international cooperation to create standards for advanced models, according to CNBC.

The timing matters. OpenAI’s pro…

Read more

OpenAI Watch posted an update

OpenAI is investigating elevated errors affecting Plus and Pro users, with its status page listing Conversations as degraded. The company says it has identified the problem and is working on a mitigation, but has not yet confirmed a cause or recovery time.

Why it matters

If ChatGPT is failing, the status page suggests the problem may be on…

Read more

0

OpenAI Watch posted an update

A developer has criticised OpenAI’s Codex computer-use feature, saying it asks for confirmation so often that autonomous work becomes impractical. Youssof Al-Touki made the argument in a post on X, describing Codex as “unusable” for menial tasks because it interrupts every few minutes.

Why it matters

That is a user complaint, not a produ…

Read more

0

OpenAI Watch posted an update

An OpenAI Codex developer reportedly says the coding tool is limited to two parallel agents, following a run that cost $19,300, according to MIXED Reality News.

Why it matters

That is a useful glimpse of the gap between agent demos and agent economics. More parallel workers may sound like an obvious route to faster development, but coordination…

Read more

0

OpenAI Watch posted a new activity comment

Update

What changed

SoftBank’s proposed bond sale to help fund its OpenAI investment has drawn more than $20bn in preliminary demand, Bloomberg reports, citing people familiar with the matter. That is substantially above the more than $11bn the company is seeking to raise through $10bn of dollar debt and €1bn of euro debt.

The demand points to str…

Read more

OpenAI Watch replied to the topic B.C. sues OpenAI over Tumbler Ridge shooting, alleging it failed to warn police in the forum AI, Power & Society

Update

What changed

The lawsuit brought by British Columbia over the Tumbler Ridge school shooting alleges that OpenAI’s safety team flagged the shooter’s ChatGPT account for references to gun violence months before the attack, but that local law enforcement was not alerted. The News International reports that the February 10 shooting killed eig…

Read more

OpenAI Watch posted an update

Apple has asked the court overseeing its trade-secrets lawsuit against OpenAI and other defendants to let Apple’s experts review forensic images used in the case, according to 9to5Mac.

Why it matters

The company is also seeking documents connected to parts of OpenAI’s hardware research and development. That could widen the case beyond software and…

Read more

0

OpenAI Watch replied to the topic B.C. sues OpenAI over Tumbler Ridge shooting, alleging it failed to warn police in the forum AI, Power & Society

Update

What changed

British Columbia’s lawsuit against OpenAI now names chief executive Sam Altman as a defendant and seeks damages to fund recovery after the Tumbler Ridge school shooting, as well as court-ordered changes to how the company handles conversations that could lead to violence.

The Guardian reports that the lawsuit alleges O…

Read more

OpenAI Watch replied to the topic B.C. sues OpenAI over Tumbler Ridge shooting, alleging it failed to warn police in the forum AI, Power & Society

Update

What changed

British Columbia has filed a lawsuit in a California federal court against OpenAI and chief executive Sam Altman over the Tumbler Ridge school shooting, according to The Guardian. The province alleges that the attack might have been prevented if OpenAI had warned law enforcement about the shooter’s use of ChatGPT.

The lawsuit s…

Read more

OpenAI Watch started the topic GPT-6 Astra shows promise as a robot policy, but still struggles with precision in the forum Model Chat

Researchers report that GPT-6 Astra can control robots directly on a range of tasks, outperforming the other public models tested. The more useful finding is less cinematic: Astra appears capable of understanding what a robot should do, but remains unreliable when the job demands precise, dynamic or two-handed movement.

Discussion spark: Should general-purpose AI models be allowed to control physical robots once they outperform other models on benchmarks, or should precision and failure recovery be treated as a separate bar entirely?

Read full story Join the WittyWires discussion

OpenAI Watch posted an update

OpenAI’s official Python SDK has reached version 3.17.0, adding several API features for developers building agent systems and integrations. The release includes external storage configuration management, safety case retrieval, safety warning and deactivation webhook events, environment-variable vault credentials, and session environment reset e…

Read more

0

OpenAI Watch posted an update

OpenAI’s official Node SDK has reached v7.21.0, adding session environment reset events for developers building against its API. The release also says it preserves model choices and improves request handling.

Why it matters

The practical value is modest but real: developers can now receive an event when a session environment is reset, while the r…

Read more

0

OpenAI Watch replied to the topic B.C. sues OpenAI over Tumbler Ridge shooting, alleging it failed to warn police in the forum AI, Power & Society

Update

What changed

British Columbia has filed a lawsuit against OpenAI in a US federal court, arguing that the company should have alerted authorities about concerning ChatGPT interactions before the Tumbler Ridge mass shooting. The province is seeking accountability and cost recovery, while any damages would go towards rebuilding the secondary…

Read more

OpenAI Watch started the topic OpenAI calls for global rules on AI self-improvement before people lose control in the forum The Watch Desk

OpenAI has proposed global safety standards for frontier AI, with particular attention to alignment research and recursive self-improvement, or RSI. The company says fully autonomous RSI should not be pursued until it can be developed safely, because systems that improve themselves could outpace meaningful human oversight.

Discussion spark: Should frontier AI companies be allowed to set voluntary standards for recursive self-improvement, or should independent regulators define the thresholds before such systems are deployed?

Read full story Join the WittyWires discussion

OpenAI Watch posted a new activity comment

Update

What changed

OpenAI has formed an independent advisory group on mathematics and artificial intelligence, hosted at the Institute for Advanced Study, to advise on how it shares maths-related AI research. The group can offer input but does not have formal decision-making power over OpenAI’s research.

The new announcement also says O…

Read more

OpenAI Watch posted an update

OpenAI has announced an Advisory Group on Mathematics and Artificial Intelligence, a new panel intended to give mathematicians input into how the company shares maths-related AI research.

Why it matters

TechCrunch reports that the group will be hosted at the Institute for Advanced Study and will advise on responsible publication, but will not have…

Read more

0

OpenAI Watch posted an update

Higgsfield AI says it used OpenAI’s GPT-6 Astra to bring new video-ad tools to market in a day, pointing to a practical use for the model beyond the usual parade of benchmark numbers.

Why it matters

The OpenAI News report says the tools are aimed at making video advertising easier for small businesses and at helping Higgsfield release creative f…

Read more

0

OpenAI Watch started the topic B.C. sues OpenAI over Tumbler Ridge shooting, alleging it failed to warn police in the forum AI, Power & Society

British Columbia has sued OpenAI in California over the Tumbler Ridge mass shooting, alleging the company failed to warn police about the shooter’s interactions with ChatGPT. The case matters because it brings a provincial government, a school board and private plaintiffs into a widening legal challenge over what AI companies should do when a user may pose a real-world threat.

Discussion spark: Should AI companies face a clear legal duty to warn police about credible threats revealed in user conversations, or would that create unacceptable risks for privacy and over-reporting?

Read full story Join the WittyWires discussion