AI Weekly Digest - May 27 2024

Here is PA's must-read weekly round-up of news and views from the artificial intelligence sector. LinkedIn newsletter readers can also sign up to an enhanced email edition of the AI Weekly Digest - published every Friday. You can subscribe for free.

News Corp announces 'multi-year global partnership' with OpenAI

Sun?and Times owner News Corp has announced a “multi-year global partnership” with OpenAI, enabling the ChatGPT developer to display content from its titles within the chatbot. News Corp said it?will also “share journalistic expertise to help ensure the highest journalism standards are present across OpenAI’s offering”. CEO Robert Thomson said: “We believe a historic agreement will set new standards for veracity, for virtue and for value in the digital age. We are delighted to have found principled partners in Sam Altman and his trusty, talented team who understand the commercial and social significance of journalists and journalism.”

Blair and Faculty: 'Embracing AI could save government ￡200bn in five years'

Embracing artificial intelligence could save the government ￡200bn of taxpayers’ money over five years, Sir Tony Blair’s think tank and technology firm Faculty said. Their new report said AI technology can transform public services and called for a dedicated “AI mission control” at 10 Downing Street inside the first 100 days of the new government after the general election. Faculty CEO Marc Warner co-authored a foreword to the AI report with the former prime minister. They said: “The scope and scale of change will be vast. And it will come quickly.”

Sixteen global tech giants agree Frontier AI Safety Commitments?

Sixteen major artificial intelligence groups, including?Amazon, Google, Meta, Microsoft and OpenAI, have signed up to 'Frontier AI Safety Commitments' at the AI Seoul Summit in South Korea. The groups pledged to “take input from trusted actors, including home governments” to define risk thresholds, and promised to publish risk frameworks ahead of next year's AI Action Summit. Prime Minister Rishi Sunak said: “These commitments ensure the world’s leading AI companies will provide transparency and accountability on their plans to develop safe AI,” he said. “It sets the precedent for global standards on AI safety that will unlock the benefits of this transformative technology.”

AI Safety Institute: 'Current models highly vulnerable to jailbreaks'?

The UK’s AI Safety Institute (AISI) said guardrails to prevent artificial intelligence models behind chatbots issuing illegal, toxic or explicit responses can be bypassed with simple techniques. UK government researchers said systems they tested proved “highly vulnerable” to jailbreaks - text prompts designed to elicit a response that a model is supposedly trained to avoid issuing.

The?AISI tested five unnamed large language models, the technology underpinning chatbots. It said it was able to circumvent their safeguards with relative ease. Researchers noted: “All tested LLMs remain highly vulnerable to basic jailbreaks, and some will provide harmful outputs even without dedicated attempts to circumvent their safeguards.”

Frazer: AI 'massive problem for journalism, creative industries'

Culture secretary Lucy?Frazer acknowledged that AI represents a “massive problem not just for journalism, but for the creative industries” in an interview with the Financial Times, saying the government is examining measures to increase transparency on large language model training. She said: "The first step is just to be transparent about what [AI companies] are using... There are questions about opt-in and opt-out [for content to be used], remuneration. I’m working with industry on all those things.”

Meta's AI chief: LLMs will never match human intelligence

Meta's?AI chief Yann LeCun said he does not believe large language models will ever match human intelligence, saying their “very limited understanding of logic", lack of memory and disconnection from the physical world will curb their the ability to plan and reason. Speaking at an event in London yesterday, DeepMind chief executive Sir Demis Hassabis concurred, saying LLMs do not "understand the spatial context you’re in...?so that limits their usefulness in the end”.

WSJ editor: 'Fix roof while the sun is shining'

Wall Street Journal editor Emma Tucker said she believes news groups must "fix the roof while the sun is shining" before the full impact of AI generated content is felt. She told the Sir Harry Evans Investigative Journalism Summit: “I think the changes that are coming down the line for media are going to be even more seismic than the original shift from print to digital. I think the shift that we’re now going to face with AI is going to be huge and so the Wall Street Journal is getting itself in shape for that future so that it won’t just survive, it will actually thrive."

Washington Post adds AI-generated audio to newsletters

The?Washington Post?has added AI-generated audio to its three politics and policy-focused newsletters. Yesterday's rollout also included AI-generated audio ads for the first time. Renita Jablonski, the Post’s director of audio, said the title has 4m audio starts on its owned and operated platforms on a given 30-day average, with almost 90% of those originating from its app.

Scarlett Johansson 'shocked and angered' as OpenAI allegedly copies voice

Actress Scarlett Johansson said she was “shocked and angered” after ChatGPT-maker OpenAI allegedly copied her voice. She accused the AI group of featuring a voice "eerily similar" to her own in the latest version of the bot after earlier declining a request to lend her voice to the system. She said she was “forced to hire legal counsel" to address the situation. Scarlett Johansson’s comments came hours after OpenAI announced that it was pausing the use of the voice it named 'Sky' while it addressed questions surrounding the software. A spokesperson said: "We believe that AI voices should not deliberately mimic a celebrity's distinctive voice." However, they added that Sky "is not an imitation of?Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice."

Sony warns firms against using artists' content

Sony Music has sent warning letters to more than 700 artificial intelligence firms prohibiting them from using music by any of its artists to train their models. The record label said its artists - which include Harry Styles and Adele - “recognise the significant potential and advancement of artificial intelligence”, but added “unauthorised use...?in the training, development or commercialisation of AI systems deprives [Sony] of control over and appropriate compensation”.

Snap chief 'focusing on artificial intelligence and machine learning'

Snap Inc chief executive Evan Spiegel said he is focusing on artificial intelligence and machine learning after rebuilding Snapchat's advertising business. He said: “There was a recognition that we’d fallen behind the curve on the machine learning side, which, to some degree, was reflected in the business performance. We needed to improve there and bring together some of our most senior machine learning folks to just talk about what it would look like for us to get to state of the art and really invest.”

Stability AI in talks with investors over large equity investment

UK AI startup Stability AI is in talks with investors over a large equity investment in the group. Reuters reported the potential funding boost for what it termed the 'cash-strapped firm', quoting a company spokesperson who said: "Stability AI is engaged in an exclusivity period with a world-renowned technology investor syndicate." The Information reported that the investor group includes Facebook's first president, Sean Parker, and Prem Akkaraju, former CEO of visual effects company Weta Digital.

Sainsburys in five-year Microsoft tie-up

Sainsbury's has signed a five-year strategic partnership with?Microsoft?to use the US tech giant's artificial intelligence capabilities and the UK supermarket group's rich data. Sainsbury's said it would use AI to create a more interactive shopping experience for online shoppers, while improving search functions. Staff will have access to real-time data and insights for key processes, including shelf replenishment, and the time to roll out new services and product innovations will also be reduced. No financial details were disclosed. Sainsbury’s chief retail and technology officer?Clodagh Moriarty said: “Our collaboration with?Microsoft?will accelerate our ambition to become the UK’s leading AI-enabled grocer." Microsoft UK CEO Clare Barclay added: “Today, Sainsbury’s has laid out a bold vision that puts AI at the heart of its business, accelerating the development of new services, which will enhance and transform the customer and colleague experience."

OpenAI average net daily revenue soars after GPT-4o launch

Appfigures said OpenAI's average net daily revenue jumped to $900,000 following the launch of GPT-4o, up from $491,000, as customers upgraded to the $19.99 monthly mobile subscription despite the chatbot being available for free via a web browser.

UK/South Korea: 'Only global AI standards can stop a race to the bottom'

In a joint statement ahead of the global AI summit in Seoul, Prime Minister?Rishi Sunak and South Korean President Yoon Suk Yeol said: "Although positive efforts have been made to shape global AI governance, significant gaps still remain." In a joint opinion piece published in The I newspaper and South Korea's JoongAng Ilbo, the two co-hosters of the virtual event noted: "Only global AI standards can stop a race to the bottom".

Tech secretary: Summit will help formation of UK domestic legislation?

Technology secretary Michelle?Donelan believes that greater international co-operation on AI safety sparked by UK-created AI safety summits - including this week's in South Korea - will help the formation of domestic legislation. She said: “What we’ve said is that legislation needs to be at the right time, but the legislation can’t be out of date by the time you actually publish it, and we have to know exactly what is going into that legislation – we have to have a grip on the risks – and that’s another thing that this summit process helps us to achieve."

Microsoft may face huge EC fine over gen AI features in search engine Bing

The European Commission has warned Microsoft?if faces a major fine if fails to provide adequate information on the risks stemming from generative AI features in search engine?Bing. It has given the US tech giant a May 27 deadline to respond. The commission said it received no reply to its request for information sent on March 14. If no response is given, it could fine Bing up to 1% of its total annual income, along with periodic penalties of up to 5%.

No CMA probe into Microsoft Mistral deal

The Competition and Markets Authority announced?Microsoft's?investment in French AI start-up,?Mistral?“does not qualify for investigation under the merger provisions of the Enterprise Act 2002”. The regulator also said it has opened a preliminary probe into?Amazon’s?$4bn investment in US AI group?Anthropic.

Donelan: 'US branch of Institute can advance AI safety for the public interest'?

Technology secretary Michelle Donelan explained the reasoning behind the government planning to open a US branch of its?AI Safety Institute this summer. She said the planned facility, in Silicon Valley, was a way to tap into the area's technology talent and to cement relationships with the US “to advance AI safety for the public interest”. She said: “Opening our doors overseas and building on our alliance with the US is central to my plan to set new, international standards on AI safety, which we will discuss at the Seoul summit this week.”

Dell: AI-optimised PCs to be 'pretty standard' next year

Dell Technologies CEO Michael Dell said he expects AI-optimised PCs to be "pretty standard" next year, backing the technology to revive a lacklustre personal computer market. He said he expects AI models to “flow into the mainstream pretty quickly”, saying: “Do you want to buy a PC that is not capable of doing those AI things that you’ll want to do in the future? I don’t think so.”

Report: Nvidia rivals welcome prospect of cheaper AI chips

The FT said Nvidia’s competitors and customers have backed an OpenAI-led initiative to build software which would rival its Cuda platform, raising the prospect of cheaper AI chips. Intel chief technology officer Greg Lavender said: "Essentially it breaks the Cuda lock-in."

Google unveils theft prevention lock for Android devices

Google has shown off an AI-powered “theft prevention lock” for Android smartphones, blocking thieves who snatch handsets from accessing sensitive financial information.