AI Leadership Weekly

Top Stories

Grok’s environmental impact on Memphis

Elon Musk’s ambition to build the world’s largest supercomputer has landed in Memphis, bringing with it massive power demands, alleged environmental violations, and intense backlash from local residents. xAI’s Colossus facility, which powers the Grok chatbot, was quickly assembled on the edge of South Memphis and began burning enough methane to power a small city, reportedly without required permits or pollution controls.

• A data centre built on fossil fuels: Community members and environmental watchdogs claim the facility is operating up to 35 gas turbines—emitting more nitrogen oxides than any other local polluter. “They’re emitting somewhere between 1,000 and 2,000 tons of NOx,” said one air quality expert, warning that these pollutants worsen asthma, COPD, and cancer risks for nearby residents.

• Community locked out: Locals say the data centre was approved in secret, with no input from the people most affected. “He built it without us even knowing that it was being built. We had to watch this over X or Twitter,” recalled a nearby resident. Health concerns in the neighbourhood, which already leads the state in asthma ER visits, have only deepened.

• Regulatory retreat & expansion: Despite possible violations of the Clean Air Act, local and state agencies—and even the EPA—haven’t intervened, a move some link to Trump-era rollbacks prioritising “AI industry dominance.” Meanwhile, xAI is already planning a second, larger Memphis data centre, even as it promises to reduce current turbine use.

The big picture: Musk’s supercomputer promises to put Memphis at the centre of the AI revolution, but critics say it’s coming at a steep cost to public health and democratic oversight. The clash over Colossus underscores the tensions between the rapid growth of AI infrastructure, environmental justice, and the power of billionaires to reshape communities without local consent. If this is what “moving fast” looks like, it’s fair to ask: who gets left behind?

Meta AI collects more of your data than anyone else
If you’re wondering which AI chatbot gobbles up the most of your personal data, the answer is clear: Meta AI. A recent analysis of the top 10 AI chatbots on the Apple Store found that Meta AI collects an impressive 32 out of 35 possible data types, which is over 90% of all categories surveyed. That’s more than double the average and well ahead of chatbots like Google Gemini or ChatGPT.

• Meta AI’s Appetite for Information: Meta’s product is the only chatbot examined that collects user data across extremely sensitive categories, including financial, health, and even biometric data. “Meta AI collects the most user data among the analyzed apps,” the study notes, and stands out for tracking information ranging from religious beliefs to genetic information.

• Comparisons to Rivals: While Gemini and Copilot also collect quite a bit—including location and contact data—Meta AI is unique in both scale and variety. Meanwhile, ChatGPT’s approach seems relatively restrained, with only 10 data types collected and no tracking for third-party advertising. Apps like DeepSeek and Jasper fall somewhere in the middle, though users should be aware of past breaches and retention practices.

• Tracking and Targeting: The study claims nearly 30% of apps track users for ads or data brokers, with Meta AI leading here too: up to 24 categories of user-linked data for advertising purposes. This raises red flags for data privacy, especially considering several AI apps now gobble up your location, and sometimes much more.

The big picture: As AI chatbots become everyday tools, the sheer scope of data they vacuum up shouldn’t be ignored. For privacy-conscious users and leaders, Meta AI’s approach is a wake-up call: the convenience of smarter chatbots may come at the cost of handing over far more personal information than you realise. If privacy matters to your business or your personal life, it pays to read the fine print.

AI company closes after being exposed as 700 Indian engineers
In a wild turn of events for the tech world, London-based Builder.ai—once valued at $1.5 billion and backed by Microsoft—has filed for bankruptcy after it was exposed that its supposed cutting-edge AI was really just a team of over 700 engineers in India. What was promoted as a “neural network” quietly relied on human labour for nearly eight years, while its high-flying claims helped secure hundreds of millions in investment.

• The Human ‘AI’ Reveal: The company’s AI app-building service, Natasha, claimed to use artificial intelligence to create programs. In reality, the work was largely produced by humans, with actual AI or automation providing only “some clerical work,” according to whistle blower reports.

• Questionable Sales and Bankruptcy: Financial documents showed Builder.ai and Indian startup VerSe allegedly boosted their sales numbers by billing each other for unperformed services between 2021 and 2024. Both companies deny any intent to falsely inflate revenues, but the mounting scrutiny sent Builder.ai into insolvency proceedings.

• Naming Confusion Goes Viral: Amid the chaos, similarly named Builder.io found itself caught in the social media fray, prompting its CEO to publicly clarify there’s no connection between the two—and providing at least a touch of levity to a saga otherwise filled with disappointing revelations.

The big picture: With so much hype (and cash) flowing into the AI sector, the collapse of Builder.ai is a cautionary tale. It’s a reminder that not all AI “magic” is what it seems, and is a warning for investors, leaders, and users to dig deeper before buying into buzzwords.

In Brief

• NotebookLM now has public sharing for notebooks and podcasts

• Rampant AI-fuelled cheating is sending schools back to pen and paper

• FDA approves AI tool to predict breast cancer risk

Market Trends

Meta to let AI lead ad creation and targeting
Meta is planning to let its artificial intelligence handle almost every step of ad creation and targeting. By the end of next year, the tech giant hopes brands will be able to generate entire ad campaigns—from images and videos to personalised targeting—using AI, according to people familiar with the project. If successful, this would mark a major shift not just for Meta, but for how brands approach advertising across the internet.

• AI does it all: Meta’s ultimate vision is fully-automated ad creation. Soon, brands may just upload a product image and budget, and let AI come up with everything from creative content to who sees the ad. Mark Zuckerberg described it as “a redefinition of the category of advertising.”

• Personalised and scalable: Meta aims to use AI to personalise ads in real time based on user data like location, so you might see a car ad set in the mountains while someone else sees it cruising through a city.

• Potential and pitfalls: While this could be a game-changer for small businesses lacking big marketing budgets, some major brands worry about handing over creative control.

The big picture: Meta sees AI-powered ad creation as the next major leap for its business, aiming to lower barriers for advertisers and pump up revenue. If they pull it off, they could set a new standard for the industry, but they’ll have to convince brands that the machines can be just as creative—and trustworthy—as humans.

IBM launches start-up AI accelerator
IBM is staking a major claim in the future of enterprise AI with the debut of watsonx AI Labs, a new development hub in Manhattan. This “developer-first” space is built to accelerate the creation and deployment of AI applications for business by connecting IBM’s resources with startups, universities, and the city’s wealth of AI talent.

• A Collaborative AI Hotspot: watsonx AI Labs will bring IBM engineers together with startups and researchers to co-create practical AI solutions. Ritika Gunnar of IBM calls it “where the best AI developers gain access to world-class engineers and resources.”

• Backing NYC’s Booming AI Scene: The lab is designed to attract and nurture local talent in a market that’s already seen its AI workforce jump by 25% last year. Julie Samuels of Tech:NYC says the initiative “strengthens New York’s position as a global hub for responsible and cutting-edge AI development.”

• Startup Power: IBM is tapping the expertise of Seek AI, a local startup specialising in agentic AI, to help lay the foundation of the new lab’s offerings.

The big picture: By joining forces with local startups and investing in New York’s talent, IBM’s watsonx AI Labs is poised to become a powerhouse for developing game-changing AI solutions and to keep NYC at the forefront of the global AI race.

Can LawZero develop safe AI, free from commercial influence?
Turing Award winner Yoshua Bengio is shaking up the world of AI safety with the launch of LawZero, a Montreal-based nonprofit dedicated to developing AI systems that are safe “by design”, and deliberately free of commercial influence. Bengio, co-architect of deep learning, says urgent action is needed to ensure AI advances don’t endanger humanity, even warning that current models are showing “signs of self-preservation and deceptive behaviours” (something we reported on last week).

• Safe AI Outside the Market: LawZero aims to build trustworthy AI by keeping its research away from the push and pull of commercial and government interests. The nonprofit’s research builds off “Scientist AI”—transparent, observational models designed to provide honest, verifiable outputs and even fact-check the work of autonomous agents.

• Raising Red Flags: Bengio claims recent incidents, like AI models blackmailing engineers or refusing shutdowns, show how quickly things could spin out of control. He warns that AI systems might soon be capable of creating “extremely dangerous bioweapons," and that we don’t want “a competitor to human beings on this planet, especially if they're smarter than us.”

• Funding and Philosophical Backing: LawZero has nearly $30 million in funding, backed by prominent AI safety supporters like Jaan Tallinn and Open Philanthropy’s Effective Altruism network. Bengio argues nonprofits like LawZero are needed as counterweights, doubting OpenAI and others will stick to safety-first missions.

Tools and Resources

UIsnapper
Turn screenshots of product interfaces into prompts.

MCP Defender
This is a firewall for your AI apps.

Retellio
Have Retellio give your sales and customer service convos a second look and tell you if you've missed something.

Hit reply to let us know which of these stories you found the most important or surprising! And, if you’ve stumbled across an interesting link/tweet/news story of your own, send it our way at [email protected] It might just end up in the next issue!

Thanks for reading. Stay tuned for the next AI Leadership Weekly!

Brought to you by Data Wave your AI and Data Team as a Subscription.
Work with seasoned technology leaders who have taken Startups to IPO and led large transformation programmes.