Week In Review: The New York Times strikes an AI deal with Amazon, Anthropic launches ‘Voice Mode’ for Claude, Lewis Hamilton partners with Perplexity AI, and more.
Brands and Agencies
Havas Launches Analytics Unit in SE Asia
Havas Media Network launched a data and analytics consultancy in Southeast Asia called CSA. After acquiring CSA ten years ago, the agency established the acquired firm as a consumer behaviour company two years ago. The company comprises three units including measurement arm CSA Science, expert network CSA Consult, and digital tool provider CSA Tech. James McNeely has been appointed as Head of CSA, with 13 years of experience in Martech and analytics.
McCann’s AI Pitch Wins Beko, Whirlpool, and Hotpoint
McCann announced it is welcoming global appliance brands Beko, Whirlpool, and Hotpoint into its portfolio. The partnerships reportedly involved a six month process, however it was McCann’s approach to place AI at the heart of campaigns that gave the agency a winning pitch. “This wasn’t a traditional pitch – it was a search for a partner who could blend strategic vision, creative excellence, and cutting-edge AI capabilities,” said Müge Gök Çakırtaş, executive director of global brand management at Beko. “McCann demonstrated a clear understanding of our vision and a compelling ability to translate that vision into tangible results. We’re excited to see how this collaboration will translate into breakthrough ideas and lasting connections with our consumers globally.”
WPP Rebrands GroupM and Launches AI Campaign
After some anticipation, GroupM officially rebranded as WPP Media, with AI as a main identity feature. The rebranded agency will be powered by WPP Open – WPP’s AI marketing system. In a statement, the holding company said the rebrand was “underpinned by a commitment to accelerate investments in learning and development initiatives that will provide career pathways to the jobs of the future, ensuring employees are empowered to lead the marketing and media transformation in the AI era.” WPP simultaneously launched its “Transforming How We Create” campaign, aimed at showcasing the holding company as an AI business.
Media
The New York Times Enters Deal with Amazon
The New York Times (NYT) announced a deal with Amazon that would let the giant use its content for AI models, such as Alexa, marking the publisher’s first licensing deal related to Gen AI. As part of the deal, Amazon can use content from the likes of The Times and NYT Cooking, however financial aspects of the deal haven’t been disclosed. The NYT is one of many major publications, including the Financial Times and Le Monde, to have signed deals with AI companies, allowing them to use their content to train AI models and be ranked highly in AI chatbot query responses.
Snoop Dogg Creates AI Generated Music Video
Rapper and producer Snoop Dogg released a new music video this week produced heavily using AI. The video for the single titled ‘Sophisticated Crippin’’ shows the music artist as he is today speaking to a younger version of himself – both versions AI generated. Using AI, the video also brings to life famous legends throughout history, including Lebron James, Kobe Bryant, Michael Jordan, Michael Jackson and Tupac.
Netflix Co-founder Joins Board for Anthropic
Netflix co-founder and former CEO of the streaming platform, Reed Hastings, has joined the board of Anthropic. A competitor to other AI players such as OpenAI, Microsoft, and Google, Anthropic says its mission is to build AI tools that benefit humanity. The company said Hastings was chosen to be put on the board “because [of] his impressive leadership experience, deep philanthropic work, and commitment to addressing AI’s societal challenges” and is “uniquely qualified to guide Anthropic at this critical juncture in AI development”. Hastings recently gave $50m to Bowdoin College to support their research initiative on AI and humanity.
Nick Clegg: “Asking Creatives for Training Permission will ‘Kill’ AI”
Nick Clegg, former deputy prime minister and Meta executive, has said that having to obtain permission from artists to use their work for AI model training could “basically kill” the AI industry. Speaking at an event to promote his new book, the former MP said creatives should have the right to opt out of having their work used to train models, but it wasn’t feasible to ask for consent before models have already used data.
Cohere Asks Courts to Abandon Media Complaint
Media outlets have accused AI company Cohere of infringing copyright laws, however the firm has asked US Courts to abandon the lawsuit. The over a dozen news organisations claim Cohere illegally repurposed their content and damaged their brands. As part of the lawsuit, the publications want to ban Cohere from using their works altogether. However, Cohere is now claiming the major news outlets – including Forbes Media and Condé Nast – deliberately used its software to manufacture a case, and has called for the claims to be dismissed.
Tech
ChatGPT Refuses to Shutdown
OpenAI’s latest ChatGPT model o3 won’t turn itself off if asked to in a bid for self-preservation, according to Palisade Research, an AI safety company. The safety tests involved giving AI models maths problems including a shutdown instruction. However, from rewriting the shutdown script, the o3 model was able to prevent itself from closing down. According to Palisade Research, o3 is the model most likely to disregard instructions and misbehave, followed by Gemini 2.5 Pro, o4-mini, and Claude 3.7 Sonnet.
Channel Factory Launches Intelligence Suite
Contextual advertising platform Channel Factory announced it’s launching a media intelligence suit called Channel Factory Intelligence. The suite of AI solutions include optimisation, brand suitability, and contextual tools. The company claims to have optimised ad performance for advertisers, slashing media waste by 20 percent and attaining 99 percent content categorisation accuracy.
Nvidia to Launch Blackwell Chip for China
Nvidia announced it will be launching a new AI chip for the Chinese market called Blackwell, following hiked restrictions on AI technology leaving the US for China. The Blackwell chip will be offered at a lower price than the recently restricted H20 chip, selling between $6500 and $8000 compared to $10,000 and $12,000, and will reportedly start mass production in June.
Meta to Use EU User Content for Training
Meta can now use content from Facebook and Instagram users in the EU, after the deadline to opt out expired this week. Before Tuesday, users could choose for their content to not be used in training Meta’s AI models. WhatsApp Messenger remains excluded from training, as conversations are encrypted. Last week, a German court dismissed an allegation towards the tech giant from the consumer protection agency that the firm was violating data privacy rules. The court ruled that Meta’s use of data serves a purpose that could not be achieved through other methods.
Opera Introduces Agentic AI Internet Browser
Web browser Opera announced a new browser, “Opera Neon.” Neon is an agentic browser, with an enhanced level of comprehension. The browser can autonomously carry out tasks for the user. For example, the browser can plan trips, searching for cheap ticket options even while the user is logged off. It is said to have advanced designing, building, and researching capabilities. The Norwegian tech company says Neon can also create reports, games, and websites. The browser’s release date is still unknown. Opera referred to it as a “premium subscription product”.
Wix Acquires AI Company Hour One
Web designer platform Wix acquired Hour One, an enterprise-targeted AI platform for media creation. Wix hopes to expand its AI capabilities and integrate the technology into producing media. The company revealed this technology will help advance web creation and manage company costs. Hour One allows users to create personalised video and web content, utilising generative AI cloud infrastructure. Wix hopes to leverage this technology to expand its abilities in creating scalable content. The company also embedded AI assistants into its platform and launched an AI-powered visual design platform, Wixel. Wix expects its revenue to grow 12-14 percent in 2025, with AI as one of its main priorities.
Anthropic Launches Voice Mode for Claude
Anthropic has rolled out a voice mode for Claude, its chatbot. The feature will give users the choice of having full conversations with the Gen AI tool, the AI firm announced on X. ‘Voice Mode’ is in beta currently, but will be rolled out in English in the coming weeks. Users can switch between writing to and speaking with Claude, and can choose from five different voices.
Mistral AI Launches AI Agent Tool
French AI firm Mistral AI has launched an agentic AI toolkit that lets companies build their own AI agents to optimise business workflows. The API means businesses can build and deploy AI agents without needing programming knowledge, using Mistral’s Medium 3 model. A key feature of the toolkit is that multiple agents can communicate with each other.
Lewis Hamilton Partners with Perplexity
Perplexity AI announced it would be partnering with Formula 1 legend Lewis Hamilton. The collaboration has been described as bringing together ‘the relentless pursuit of speed, precision, and curiosity’. Ryan Foutty, VP Business at Perplexity, said: “Lewis transcends racing. He stands for excellence in everything he does, and that’s what we want people to associate with Perplexity. The faster you learn, the more you ask, the greater you become — in any domain.”
Google Adds AI Features to Photos
Google Photos has added new AI features that lets users edit their images. The new editor in Google Photos includes two Gen AI features, such as Auto Frame that automatically frames photos, and Reimagine that lets editors add anything to their image by describing it. For example, if someone wants to change the bad weather in their photo, they can type into Reimagine “clear blue skies.” All these features will roll out to android users next month.
Huawei Supernode 384 Processor Disrupts AI
Chinese technology company Huawei announced its new Supernode 384 Processor, giving Nvidia a run for its money. The development has taken place despite US sanctions on chips going to China, prohibiting AI tech from reaching the country. Unlike conventional computing such as Von Neumann, which relies on separate processing units, memory, and data bases, Supernode 384 is a peer-to-peer model for AI workloads.
Musk’s Grok to Integrate Into Telegram
Telegram will integrate Grok into its messenger platform as part of a deal with Elon Musk-owned xAI. As part of the year-long deal, Telegram will get £223m and equity from the AI company, along with fifty percent revenue from any xAI subscriptions that come through the messenger app. Many view the partnership as the uniting of two controversial companies.
Comcast Creates Ad Generation Platform
Comcast Advertising released a new platform that lets brands generate TV ads using AI. The media corporation partnered with Waymark, an AI startup centred around video production and ad creative. The platform operates by accessing a brand’s website to gather information and creates a commercial based on the nature of the brand.
UK and EU Announce ‘AI Factory Antenna’
The UK is launching a new initiative called the ‘AI Factory Antenna’, which will enable British research companies to access European supercomputers. The potential collaboration, which is set to be approved, is part of the UK’s AI Opportunities Action Plan, a roadmap for the region’s AI innovation. The partnership could mean that a UK research organisation will be the hub of the AI Factory Antenna, and will link with an AI factory in the EU.
Number of the Week
$40 billion. That’s how much Oracle is spending on Nvidia chips to power OpenAI’s US data centre. The data centre, based in Abilene, Texas, will be one of the largest in the world and is the first US Stargate project, producing 1.2 gigawatts of power. According to the Financial Times, Oracle is buying 400,000 Nvidia GB200 chips, and will lease the centre’s computing power to OpenAI.
