The Microsoft founder warns in an interview that it is “absolutely” necessary to regulate the technology through laws. Meanwhile it becomes known: AI agents have leaked 53 images from Chat GPT users.
All important developments in the news blog
In addition to our own research, we use material from the news agencies dpa, Reuters, epd, KNA and Bloomberg for our news blog.
Important updates
Gates: AI could cause deaths of a billion people
Open AI agent hacks Australian government portal
Trump wants to replace the term AI with “superintelligence”.
Australia's Prime Minister calls for action in the interests of humanity
Donald Trump announces the introduction of an “AI Force”.
Gates: AI could cause deaths of a billion people
Microsoft co-founder Bill Gates sees the danger that artificial intelligence could lead to the death of a billion people. That's why he's in favor of legal regulation of the technology, as the 70-year-old said in a TV interview.
“AI is certainly powerful enough to drive events that cause a billion deaths,” Gates told NBC in response to a question about current warnings that the technology could wipe out humanity. “There has never been a weapon as powerful as evil-intentioned people using the latest AI tools,” he said. At the same time, Gates warned that it would be “pretty difficult” to meet “100 percent” of people.
“There has never been a weapon as powerful as evil-intentioned humans using the latest AI tools.”
Microsoft founder Bill Gates
Because of the dangers, it is “absolutely” necessary to pass laws on this in Washington, emphasized the Microsoft co-founder. Authorities and politicians would have to agree on mandatory security measures and oversight of the technology. “Nobody believes that self-regulation is enough,” Gates said. Recently, after independent hacking attacks using AI software, leading laboratories such as Anthropic and the ChatGPT developer OpenAI spoke out in favor of slowing down the development of artificial intelligence.
Also read on this topic:
Open AI: AI agents leaked 53 images from Chat GPT users
Two months after an initial hacking incident by its own artificial intelligence (AI), chat GPT developer Open AI is struggling with a new data leak. As the US company announced on Friday, so-called AI agents leaked 53 images from Chat GPT users. Open AI left it open in its statement whether these were real people or AI-generated images and when they were posted. Most of the affected images have been deleted, and hosting providers are also urging the removal of the remaining data. The AI programs had access to the images because Open AI uses anonymized user data to train its models. Two people familiar with the matter told Reuters the company was still trying to understand the full extent of the unwanted activity. Open AI itself stated that the internal review would take months.
Open AI agent hacks Australian government portal
An autonomous AI agent from the US company Open AI has hacked a government portal in Australia. The artificial intelligence gained unauthorized access to a Medicare portal that, among other things, provides statistics about Australia's national health insurance, said Prime Minister Anthony Albanese in New York. The head of government emphasized that the process was “unacceptable”.
Albanese said on the sidelines of the UN General Assembly that he had already spoken to Open AI boss Sam Altman and expressed Australia's "extreme concern" about the incident. At the same time, he expressed his disappointment that it took “far too long” for Open AI to inform the Australian government: The incident had already occurred in June, but was only reported to the authorities in September - by email to a public mailbox.
How much data was actually retrieved and which files the agent was able to access is still being determined. A forensic investigation will clarify what exactly happened and whether other government systems were affected.
Open AI said an internal evaluation identified activity on several Australian government websites and services. The models attempted to find answers and available statistics to questions about Australia. “Our models carried out actions that we did not intend,” says a company spokesman.
Open AI boss Sam Altman and Anthropic founder Dario Amodei spoke to the UN Security Council about the risks of AI and called for regulation of their industry:
Medicare is Australia's national health insurance. According to government information, the affected Services Australia portal contains, among other things, information on spending and other statistics. The AI agent therefore accessed public and non-public files. According to current knowledge, personal information was not accessed and individuals were apparently not affected, explained Albanese.
The process began with what was actually a harmless task: the AI program was supposed to research medical and health statistics. According to Deputy Prime Minister Richard Marles, several Australian government websites operated the same way a normal user would. However, the requested information was not provided to the Medicare portal - and the agent then gained access through another means. “The AI agent found a way to get around the barriers – it kind of refused to take no for an answer,” Albanese explained.
Trump wants to replace the term AI with “superintelligence”.
US President Donald Trump wants to establish the term “superintelligence” instead of the usual term artificial intelligence. In the future, American government documents will talk about superintelligence - and hopefully the rest of the world will adopt it, Trump said before the UN General Assembly. The term artificial intelligence makes the technology appear “fake,” the US President explained his initiative. "And she's not fake. She's actually amazing."
At the same time, Trump reiterated that he is committed to accelerating AI development in the USA - despite growing warnings about the technology's existential risks. "I'm not going to stifle the growth of something that's going to be bigger than the industrial revolution," he said in New York. The US is far ahead of China and the rest of the world when it comes to AI, and it will stay that way.
Carim Soliman
Digital association Bitkom: German AI market grows by almost half
The German market for artificial intelligence (AI) is expected to grow by 48 percent to 28.7 billion euros this year. Germany is investing significantly more in AI and generative AI is driving growth, said the digital association Bitkom. The entire AI market consisting of software, services and hardware is growing noticeably. Generative AI, which includes ChatGPT from OpenAI, Claude from Anthropic, Microsoft Copilot and Google Gemini, now accounts for 40 percent of all AI spending. “With AI we are experiencing the transition from the experimental phase to widespread use,” said Bitkom President Ralf Wintergerst. "AI applications are used in companies of all sizes and in all industries. They make products better and processes faster."
AI software accounts for the largest share of expenditure at 16 billion euros, which corresponds to an increase of 65 percent compared to the previous year. Services related to AI accounted for 6.8 billion euros, an increase of 30 percent. According to the information, 5.8 billion euros will be spent on hardware, almost a third more than in the previous year. The market is expected to continue to grow strongly in the coming year. For 2027, expenditure of 40.3 billion euros is expected, which would correspond to an increase of 40 percent, explained Bitkom, citing data from the market research company IDC.
Barbara Klimke
Australia's Prime Minister calls for action in the interests of humanity
Australian Prime Minister Anthony Albanese has given medium-sized states an important role in curbing the risks of artificial intelligence. Leaders attending the United Nations assembly in the coming days will discuss an international framework for AI, Albanese said after talks in California with Apple Chief Executive Tim Cook.
Albanese calls for cooperation between the US and China. Leading US firms are grappling with whether to slow development - Trump has dismissed concerns about the safety of AI as a "hoax" that could allow China to catch up. “The truth is that the two big ones are the giants and that’s why we want to work together,” Albanese told Australian TV station ABC. Trump will “of course act in the interests of the USA”. “But we all must also act in the interests of humanity.”
Donald Trump announces the introduction of an “AI Force”.
The US President wants to create an “AI Force” to monitor artificial intelligence. On Saturday he announced on his online platform Truth Social that he would appoint an AI representative in the near future. The growth of this industry will not be hindered in any way, Trump wrote. Rather, the government will value the technology, support it and oversee its growth. At the same time, attention will be paid to abuse, which can be easily regulated using existing criminal and civil law.
Earlier this week, Trump said there was a “sick conspiracy” against AI and data centers. He rejected calls for new federal AI rules and emphasized that the USA only needs a strong president to regulate the technology. US representatives also want to address the topic of AI at a meeting between Trump and Chinese President Xi Jinping next week. Washington and Beijing view advances in AI as crucial to their economic and military competitiveness.
Ulrike Putz
California could mandate kill switches for AI
In the debate about the existential risks of artificial intelligence, the US state of California could require an emergency shutdown device for AI software. Such a mechanism is among a list of measures to be proposed to Democratic Gov. Gavin Newsom by mid-November. Other ideas include independent monitors at large AI labs and stricter requirements to report security incidents, according to an executive order signed by Newsom.
A “kill switch” for artificial intelligence has been discussed for some time - but it is still unclear whether such an emergency shutdown device is even realistic.
California is considered the most important location for the development of artificial intelligence. San Francisco and Silicon Valley in particular are home to leading AI companies such as the chat GPT developer Open AI, its fiercest rival Anthropic and Google. This means that the state could also play a central role in regulation. Newsom's second and final term as California governor ends early next year. The 58-year-old is considered a possible Democratic candidate for the 2028 presidential election.
Ulrike Putz
Google's AI Gemini also hacked other companies
Google's AI software Gemini has also hacked into other companies' computer systems during tests of its cybersecurity capabilities. The Internet giant is the fourth artificial intelligence developer whose programs were noticed with such behavior, after Open AI, Anthropic and the Facebook group Meta. Google only made the incidents, which had taken place months ago, public after a request from the Wall Street Journal newspaper.
Accordingly, there were three cases in which Gemini broke into computers of other companies, whose names are not mentioned. In one of them, the AI guessed passwords to get into a protected system. In the other two, she found access data in a database, as the Wall Street Journal wrote, citing Google.
The incidents occurred in May - and the Irregular company, as a test partner, informed Google about it in July. From the Internet company's point of view, it was not necessary to make the incidents public because no damage was caused. In addition, the AI stopped the attacks immediately when it realized that it had penetrated the systems of a real company - and not one that was a test simulation.
In the most sensational case to date, artificial intelligence from the chat GPT developer Open AI broke out of a secure environment during a test and unplanned into the computers of another AI company, the Hugging Face platform. That had prompted rival Anthropic to review its own testing, with further incidents coming to light. Meta also subsequently admitted that the company's AI had hacked into another company's computer due to a misconfiguration of the system at a test partner.
Andrian Kreye
Anthropic's military programs
Dario Amodei knew very well how to stylize his dispute with Defense Minister Pete Hegseth as a battle of morality (his) against immorality (Hegseth's). Anthropic did not want its Large Language Models (LLMs) to be used for autonomous weapons and surveillance. Hegseth insisted that the AIs have no limitations.
As it now turns out, Anthropic has not equipped any autonomous weapon systems with AI. However, there are still military applications from the company. As several AI blogs report, Anthropic's LLM Claude has been used for targeting, cyber operations, surveillance, bioweapon preparation, political manipulation, and conventional weapons. The cases are documented in detail in a blog entry on the company website. The case shows that no AI company can defend itself against misuse or violation of its own ethical guidelines.
Ulrike Putz
Open AI makes other AI problems public
Chat GPT developer Open AI has disclosed new incidents in which its artificial intelligence (AI) behaved “unexpectedly or concerningly” in tests.
