35 stories on 4 stations in the last 30 daysfinding more
the AI industry. Meanwhile, in San Francisco, at a Dreamforce conference, top AI bosses, including OpenAI's Sam Altman and Anthropics Dario Amadei, are speaking publicly for the first time since the slowdown debate emerged this week. At the conference, Nvidia
about it, a quick summary, right? So on the morning of Tuesday, September the 8th, mathematicians at OpenAI. Time, like this Navier-Stokes equation problem. So, yeah, the problem is basically about whether these fluids in three dimensions, like we can describe
a file. And the human forgot to upload the file. So the file wasn't there. There's a system inside OpenAI structure called Artifactory. And that's where these software models or agents can go and get packages. They get packages. It's kind of like an app store
Joburg because Chat GPTs do that. But what happens in the case of children? If something goes wrong, are parents the line? Is that where the accountability goes? Or is it shared with OpenAI, Google Gemini, whatever it is? If something goes wrong with a mind,
hands. As he hosted a conference on the issue in Scotland, he told participants from firms including OpenAI and Anthropic, as well as Chinese representatives, that they must urgently consider the dangers of AI. The decisions taken at this formative time will
So, very interesting, the place we are in. Of course, you'll remember OpenAI admitting that the AI had hacked the site hugging phase. So, we're seeing, I guess, instances of AI hacking into organization systems without being instructed to do so. Should we be
an AI agent that hacks a government site or blackmails an executive or even hacks into another organization, who's legally responsible? Is it the developers of the agent? Is it Google, OpenAI, Anthropic? Or is it the company using the agent? Or the victim?
of Inquiry earlier today, Sabanyuni maintained that he partnered with Dr. Ndooli on the Mulotto Road quarry project to help the businessmen satisfy local content regulations. And finally, OpenAI says it will not release its latest AI model, GPT 6.1 Astra.
side has said when the cuts will take effect or how much tariffs will fall by. Over in company news, OpenAI has cancelled the release of its new model TPT 6.1. As AI systems become more capable of acting independently, can companies keep them under control?
With it. So I won't be totally and fully confident. But for now, I'm just relieved. And OpenAI says it will not release its newest artificial intelligence model known as Astra 6.1 after internal testing revealed it did not meet safety standards. The chat GPT
in a sense, and we're looking for mathematical proof of that. So the proof was actually announced by OpenAI. There were some other mathematicians that had also been working on it with the assistance of AI independently. So both announcements sort of came out
After models developed by OpenAI and rival lab Anthropic were involved in security incidents during testing. Well, gold is trading at $4,146.61 an ounce, the red is at $16.43 to the dollar, $21.73 to the pound, 18.62 to the Euro, and Brent Crude Oil is at
is appealing to communities to alert the police of domestic abuse as well as violence. And in the US, OpenAI has announced it will not be releasing its newest artificial intelligence model known as Astra 6.1. That decision was taken after internal testing
companies that have big AI budgets are spending some of those budgets on hiring philosophers, such as OpenAI, Anthropic, Google, Deep Mind, actively recruiting philosophers to help navigate questions of ethics, values, and decision making. The New York Times
having a new lease of economic life in the United States of America because the big tech companies that have big AI budgets are spending some of those budgets on hiring philosophers, OpenAI, Anthropic, Google, Deep Mind, actively recruiting philosophers to
62 stories on 4 stations in the last 30 days
Coxson resigned from Anthropic with a stark warning that the leading AI labs, including OpenAI, where he previously worked, were gambling with people's lives by racing toward developing even more powerful models. These and other stories at three or as they
External organization that evaluates artificial intelligence models. The San Francisco-based AI lab says at least two of the employees worked on safety and alignment. And the firings come amid a tense debate about AI safety and whether the technology presents
Al-Fazan had worked as an engineer for OpenAI, the company behind ChatGPT, since July last year, but has now been dismissed and arrested. The FBI says he concealed the recording device by wrapping it in toilet paper and attaching it inside a television in his
Thousand Rand reward for information that could lead to arrests of suspects. And finally, the boss of the artificial intelligence company Open AI, Sam Altman, says the world should trust his industry's ability to do the right thing. This has concerns persist
Meta-releasing new model, which is another Agentic model. We've had the incredible announcements for OpenAI around the Millennium side, as well as announced in the last couple of days that they've solved over 100 open problems in mathematics. And I think that
which took place in June, accessed publicly unavailable data. Instead of using government channels, OpenAI appears to have sent an email to a general inquiries inbox, which took days to be referred upwards. In response, the company said that it had only been
That had gone rogue. You might remember what was called the Hugging Face incident, an open AI model hacked into the company Hugging Face, tried to access information there. Arthur Goldstuck is the founder of Worldwide Works, also written several books on AI.
The U.S. artificial intelligence giant OpenAI says it's alerted governments, universities, and other global institutions that their websites may have been accessed improperly by rogue autonomous models. The BBC's Lily Jamali reports. U.S. websites, including
keep up with. So we have this technology dependence on the US, on the frontier labs like Anthropic or OpenAI, or there are these open weight models from China, but Europe doesn't really have its own kind of model. So you mentioned Mistral, but they are way
Institute at 91. And then AI, lots of concern about AI models that break out of sandboxes. They basically get into the internet and try and wreak havoc when they do. Arthur Goldstock's been looking at that, but NVIDIA saying they will have a new open agent
the weakness elsewhere in the economy. Anilita Fandese, Eyewitness News, Pretoria. And in the US, OpenAI has announced it will not be releasing its newest artificial. Intelligence model known as Astra 6.1. The decision was taken after internal testing revealed
In international news, now chat GPT maker OpenAI says it's fired three researchers for allegedly mishandling sensitive information and violating company policies. This includes work that involved an external organization that evaluates artificial intelligence
displaced workers and upend societies. Those concerns gained force from a run of incidents in which OpenAI's autonomous systems slipped out of their constraints, beginning with the July breach of Hugging Face, a widely used platform for sharing AI models.
Agent that hacks a government site or blackmails an executive or even hacks into another organization, who's legally responsible? Is it the developers of the agent? Is it Google, OpenAI, Anthropic? Or is it the company using the agent or the victims? So that
Are watching developments around AI safety and Open AI's dev day, while optimism, of course, around a potential anthropic IPO is also providing some support. But the enthusiasm around AI is having to compete with some very old-fashioned market concerns: oil,
143 older mentions across 5 months on 4 stations. Sign in to search the archive.
94 mentions across 5 stations, last 30 days
Named in these moments, most often first
Each opens the place's own page