Eric Schmidt, former Google CEO has expressed concerns about the extreme risks posed by artificial intelligence (AI) falling into the hands of terrorists or rogue states.

Eric Schmidt, former Google CEO

He warned that nations such as North Korea, Iran, and Russia could adopt AI technologies to develop weapons capable of causing significant harm, including biological weapons.

Schmidt urged governments to oversee private tech companies, emphasising, “The real fears I have are not the ones most people discuss about AI, I talk about extreme risk.”

“I’m always worried about an ‘Osama Bin Laden’ scenario, where truly evil individuals take control of some aspect of modern life to harm innocent people,” he added.

With private companies driving AI advancements, he stressed the need for careful government monitoring and regulation. “It’s really important that governments understand what we’re doing and keep their eye on us,” he said.

His remarks followed a two-day AI summit in Paris, where the UK and the U.S. declined to sign a communiqué outlining the future direction of AI. The declaration on “inclusive and sustainable artificial intelligence for people and the planet” was endorsed by 57 countries, including India, China, the Vatican, the EU, and the African Union Commission.

The UK justified its decision, stating that the agreement lacked “practical clarity” on global AI governance and national security concerns.

Schmidt supports U.S. export controls restricting the sale of advanced AI microchips to certain countries, aiming to slow adversaries’ progress in AI research.

He also highlights the importance of international collaboration on AI safety, suggesting that cooperation with nations like China is essential to addressing global AI challenges.

Sam Altman, chief executive of ChatGPT-owner OpenAI, has firmly declared the company “not for sale” following a $97.4bn (£78.4bn) takeover bid from a consortium led by Elon Musk.

Elon Musk

Speaking at the AI Action Summit in Paris, Altman emphasised OpenAI’s mission to develop AGI (artificial general intelligence) for the benefit of humanity.

Marc Toberoff, attorney for Elon Musk, confirmed the bid submission on Monday.

In response, Altman humorously offered to buy Twitter for $9.74 billion on Musk’s platform.

Unlike many tech giants, OpenAI is not publicly traded but operates through a complex partnership between non-profit and for-profit entities.

Musk aims to return OpenAI to its non-profit roots, despite owning a rival firm, xAI.

Christie Pitts, a tech investor, expressed scepticism about Musk’s intentions, noting his competitive interests.

Altman echoed this sentiment, suggesting Musk’s move disregards OpenAI’s mission.

Altman, who holds no stock in OpenAI, advocates transforming the organisation into a fully for-profit company to raise more funds for AI research.

Although the board has the final say, the $97.4bn offer falls short of OpenAI’s previous $157bn valuation and rumoured $300bn in future funding talks.

Toberoff stated the consortium might increase their bid. Meanwhile, OpenAI is collaborating with Oracle, a Japanese investment firm, and an Emirati sovereign wealth fund on “The Stargate Project,” a $500 billion AI infrastructure initiative announced by President Donald Trump.

Adobe yesterday released the first public version of an artificial intelligence tool that can generate video clips and revealed how much it will charge, but said it will not set pricing for major users such as studios until later this year.

The Firefly Video Model, as Adobe is calling the service, will compete against Sora, a model developed by ChatGPT creator OpenAI, and startup Runway, both of which currently offer video-generation services. Facebook owner Meta Platforms has also developed a video-generation AI model but has not given a timeline for when it will be released.

Adobe’s model differs from its rivals because it is geared toward generating clips that will fit into how film and television studios use Premiere Pro, its flagship video editing software.

To that end, many of the features that Adobe is emphasizing revolve around feeding existing shots into the video model and asking it to generate clips that fix or expand on shots that were taken on a real production set but that did not come out quite right.

Adobe said the service will generate five-second clips at 1080p resolution. While that is shorter than the clips of up to 20 seconds generated by OpenAI’s service, Adobe executives said the majority of individual clips in most productions are only three seconds.

Adobe said a user can generate 20 clips per month for $9.99 and 70 clips for $29.99. That compares with 50 videos for $20 per month with OpenAI’s plan at lower resolution and a $200 OpenAI plan that can handle longer, higher resolution videos.

Adobe is also working on a “Premium” pricing plan for studios and other high-volume video users and will release those pricing details later this year. Alexandru Costin, Adobe’s vice president of generative AI, said the company is working to generate 4K video and will remain focused on quality rather than longer clips.

“We actually think that great motion, great structure, great definition scheme, making the actual clip look like it was film, is more important than making a longer clip that’s unusable,” Costin told Reuters.