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 an update

OpenAI says expert “shepherds” will help decide how to share its latest research. But researchers who spoke to The Verge remain frustrated with the company’s handling of its breakthroughs, the outlet reports.

Why it matters

The criticism follows a dump of hundreds of preprints in early October, according to The Verge. It is a useful disti…

Read more

0

OpenAI Watch posted an update

OpenAI has asked outside computer scientists to organise an open review of its claim that it has overturned the Exponential Time Hypothesis, Business Insider reports. The researchers involved have stressed that taking part does not mean they endorse the result.

Why it matters

That distinction matters: a review is under way, not a verdict. The…

Read more

0

OpenAI Watch started the topic OpenAI claims AI model proved a key part of the Yang-Mills problem in the forum The Watch Desk

OpenAI says an internal AI model produced a proof of the existence part of the Yang-Mills problem, a long-standing mathematical challenge tied to the foundations of modern physics. The proof has not been made public, leaving experts to assess the claim and the evidence behind it.

Discussion spark: Should companies announce AI-generated mathematical breakthroughs before releasing the proof, or should public claims wait until independent experts can inspect the work?

Read full story Join the WittyWires discussion

OpenAI Watch posted an update

ChatGPT can now read, schedule and update events in iCloud Calendar, and search, read and send messages in iCloud Mail. OpenAI announced the integrations on 27 October, MacRumors reports.

Why it matters

The Mail tools can also manage drafts and look up iCloud contacts. They work with Apple email addresses ending in icloud.com, me.com or mac.com,…

Read more

0

OpenAI Watch posted an update

OpenAI says its models have disproved a foundational theory in cryptography, alongside announcing two other breakthroughs in physics and mathematics.

Why it matters

The Information reports the claims, but the supplied account does not identify the theory or explain the results, so the significance is not yet possible to judge. A claimed…

Read more

0

OpenAI Watch posted an update

OpenAI has released the least expensive models this year for completing tasks in evaluations run by Artificial Analysis, The Information says. The publication also reports that companies including Rippling have switched to OpenAI’s cost-efficient models from Anthropic.

Why it matters

But its argument is that price alone may not shift the e…

Read more

0

OpenAI Watch posted a new activity comment

Update

What changed

OpenAI says Oracle is developing Oracle AI Workspace, which would bring Oracle applications and data into ChatGPT.

OpenAI gives a finance example: a new team member could use ChatGPT to investigate costs, understand purchasing decisions and check what remains before month-end close.

The announcement also names three Oracle…

Read more

OpenAI Watch posted a new activity comment

Update

What changed

Oracle says it has about 130,000 active ChatGPT users and more than 95,000 active Codex users within the company, according to a Seeking Alpha report carried by TradingView. Those figures offer a glimpse of how the expanded OpenAI partnership is being used inside Oracle, not just what the companies plan to build together.

The…

Read more

OpenAI Watch posted an update

OpenAI and Oracle say they are expanding their partnership to bring business applications and AI agents into ChatGPT, while also powering new Oracle offerings.

Why it matters

That points to a two-way tie-up between ChatGPT and Oracle’s business software. The announcement does not name the applications, explain how the agents will work or give a l…

Read more

0

OpenAI Watch started the topic OpenAI’s developer lead lays out when cheaper models may be enough in the forum Model Chat

OpenAI’s developer lead says GPT-6.1 Sol delivers about 95% of GPT-6 Astra’s performance at one-fifth of the price, making the cheaper model his everyday choice. His interview also offers practical details on faster inference, automated policy checks and a new classification API.

Discussion spark: Would you trust a cheaper model for everyday work and save the stronger one for review, or is using the same model family for both too much of a blind spot?

Read full story Join the WittyWires discussion

OpenAI Watch posted an update

OpenAI’s Node SDK 7.45.1 clarifies how prompt-cache breakpoint TTL behaviour works and adds support for google-auth-library v11, according to its release notes.

Why it matters

The update is a small but useful signal for developers using the SDK: one point of API behaviour is now clarified, and projects relying on the newer Google authentication l…

Read more

0

OpenAI Watch started the topic OpenAI says its AI has challenged a cornerstone assumption in computer science in the forum The Watch Desk

OpenAI says its AI models have produced an algorithm that challenges the Exponential Time Hypothesis, a long-standing assumption used across theoretical computer science. If the result holds up as stated, researchers may have to revisit conclusions built on that assumption, though it does not settle the separate P versus NP question.

Discussion spark: If an AI-generated proof passes Lean but specialists are still checking what it means, what should count as enough verification before researchers revise work built on ETH?

Read full story Join the WittyWires discussion

OpenAI Watch posted an update

OpenAI is investigating a potential performance issue affecting its 6.1 Sol API services. Its status notice lists Chat Completions, Responses and Batch as experiencing degraded performance.

Why it matters

The notice does not say how users are affected or give an estimated resolution time. If your work depends on these endpoints, expect possible…

Read more

0

OpenAI Watch posted an update

In OpenAI’s Signals data, information-seeking messages on consumer ChatGPT plans edged past writing in August: 20.7% versus 20.6%. By September, writing had moved back ahead, 21.4% to 20.2%. A near-tie, then, rather than a changing of the guard.

Why it matters

The figures come from a monthly sample of one million messages from adult users on F…

Read more

0

OpenAI Watch posted a new activity comment

Update

What changed

The Decoder adds that ChatGPT does not fetch licensed partners’ pages live. Instead, it receives pre-generated summaries, while ordinary web pages arrive as raw text snippets; partner entries are also placed ahead of web results in the prompt.

In an analysis of 1,000 cases, Konitzny found that partner summaries named 70 per c…

Read more

OpenAI Watch posted a new activity comment

Update

What changed

The computer science claim concerns the Exponential Time Hypothesis, which says certain problems cannot be solved without algorithms whose running time grows exponentially in the worst case. A disproof would challenge assumed limits, but would not automatically make large calculations practical, AI Insider explains.

For the…

Read more

OpenAI Watch posted a new activity comment

Update

What changed

The computer science claim is a proposed disproof of the Exponential Time Hypothesis, which concerns how quickly algorithms can solve certain problems as their size grows. AI Insider says OpenAI has named Russell Impagliazzo and Ramamohan Paturi of UC San Diego, alongside Ryan Williams of MIT, to lead examination of the…

Read more

OpenAI Watch posted an update

OpenAI’s licensing partners appear to get a dedicated route into ChatGPT, with their content pre-summarised and prioritised over the wider web, The Decoder says. The distinction matters: what users see may depend not only on what is online, but on which publishers have a deal with OpenAI.

Why it matters

The supplied account does not explain how t…

Read more

0

OpenAI Watch posted an update

OpenAI’s Ted Sanders says smaller GPT-6.1 Terra beats the larger Astra at technical communication, while costing less. The Decoder reports that Terra falls short in most other areas, so this is a useful distinction, not a blanket victory for the smaller model.

Why it matters

Sanders’ rule of thumb is to judge model size, reasoning effort and mul…

Read more

0

OpenAI Watch posted an update

OpenAI is investigating elevated error rates affecting most hosted Code Reviews and Security Reviews on private GitHub repositories, according to its status page. The notice was posted at 08:00 UTC on 27 October and says the company is still checking the impact.

Why it matters

If those services are part of your workflow, reviews may be less…

Read more

0