Google Fined, Intel Backed, AI Antibodies Designed
21 March 2024: $270M Google fine, $20B White House deal for Intel, UW designs 100 AI antibodies, breast cancer treatment with 73.4% accuracy, AI Deals & IPOs.
Good Morning,
Latest in AI,
1-Minute Roundup
Google fined €250 million ($270M) in France for using copyrighted news content to train an AI model without permission.
The White House is backing Intel with a $20B deal—$8.5B in grants and $11B in loans—to boost US chip manufacturing.
AI deals: Carlsmed ($52.5M), Euno ($6.25M), Buzz Solutions ($5M), Liquidnitro Games ($5.25M), Flowpay (€2.1M), Lumino ($2.8M), PocketHealth ($33M), Visio.ai ($2.5M), Graswald.ai ($3.3M), Databricks acquired Lilac, StanhopeAI (£2.3M).
OpenAI is set to launch the GPT-5 AI model this summer.
No sanctions for Trump's ex-lawyer Michael Cohen over mistakenly filing AI-generated fake case citations.
Google DeepMind's TacticAI suggestions for football positioning were favoured 90% of the time.
UW researchers used AI to design 100 antibodies targeting SARS-CoV-2.
Google AI now predicts floods in 80 countries, giving 7-day warnings to 460M people.
AI tool predicts side effects of breast cancer treatment with 73.4% accuracy.
Stability AI saw lawsuits and resignations after failing to raise $100M in 2022.
Suspicions of AI-generated Amazon cookbook with inconsistencies, fake author bio.
Astera Labs IPO surged 72%, and Reddit's upcoming IPO shows AI companies' investor appeal.
Headlines
Google Fined $270M in France for AI Model Training with News Content
France's Autorité de la Concurrence has fined Google €250 million ($270M) for using copyrighted news content to train its AI model, Bard/Gemini, without proper notification. This action adds to previous disputes over copyright and fair payment negotiations with news publishers.
Copyright Clash: Google's fine stems from unauthorised use of news content to train its Bard/Gemini AI, breaching prior commitments to notify publishers and ensure fair payment discussions.
Past and Present Penalties: This is not Google's first run-in with French authorities; it was fined $592 million in 2021 for similar copyright negotiation issues. Google has agreed not to contest the latest fine, seeking to move past the ongoing saga.
Technical and Transparency Issues: Until September 2023, publishers lacked a specific opt-out for AI training without impacting their presence on Google's other services.
White House Backs Intel with $20B for U.S. Chip Manufacturing Boost
White House has announced a $20B deal that includes $8.5 billion in grants and $11 billion in loans under the CHIPS Act. This move aims to fortify US manufacturing capabilities amidst supply chain challenges and geopolitical tensions, particularly with China.
Strategic Funding: The proposed $8.5 billion support for Intel is part of broader efforts to strengthen US chip manufacturing amidst global supply chain disruptions and competitive pressures.
Intel's Commitment: Intel plans to invest tenfold over the next five years in expanding its manufacturing footprint across Arizona, New Mexico, and Oregon and in a new $10 billion facility in Ohio.
Long-term Impact: Challenges like competition from global giants like TSMC and operational delays in projects like the Ohio plant raise questions about the immediate impact.
AI Deals
Carlsmed, an AI Medical Technology company, secured $52.5M in Series C financing led by B Capital and US Venture Partners. The company aims to expand its personalised spine surgery platform with cervical fusions.
Euno raised $6.25M in seed funding to launch its AI-enhanced data model governance platform. It bridges the gap between data teams and business analysts for seamless, autonomous, and aligned data handling.
Buzz Solutions has raised $5M, led by GoPoint Ventures, to expand its AI-powered grid infrastructure inspection and monitoring platform.
Liquidnitro Games secured $5.25M in seed funding led by Nexus Venture Partners. The company aims to harness AI and technology to enhance game live services and production.
Flowpay, which specialises in predictive AI and automation for SME financing, has raised €2.1M in a seed round led by Techstars, Soulmates Ventures, and DEPO Ventures.
Lumino raised $2.8M in Pre-Seed funding to develop an AI infrastructure protocol that reduces compute costs by up to 80%, addressing GPU scarcity and centralisation issues in AI model training.
PocketHealth, an AI health startup, raised $33M from Deloitte Ventures and Samsung Next. It plans to use the funds to enhance its image-sharing platform and expand its healthcare network.
Visio.ai, a Brazilian startup specialising in AI and computer vision for the food and retail sector secured $2.5M in a seed round led by DGF Investments.
Graswald.ai raised $3.3M to enhance its 3D content creation platform. With AI-driven tools, it aims to revolutionise enterprise 3D content.
Databricks acquired Boston-based startup Lilac to enhance data quality for large language model (LLM) applications.
StanhopeAI has raised £2.3M for its neuroscience-inspired AI capable of autonomous drone decision-making.
Others
OpenAI Set to Launch Powerful GPT-5 AI Model This Summer
OpenAI is expected to release GPT-5, the next iteration of its large language model, in the summer of 2024.
The training data for GPT-5 reportedly includes copyrighted material, which may raise regulatory concerns around fair use.
No Sanctions for Michael Cohen Over Fake AI-Generated Case Citations
Former Trump lawyer Michael Cohen mistakenly submitted fake case citations generated by Google's AI chatbot Bard/Gemini in a legal filing but will not face sanctions.
The judge called the incident "embarrassing" but found no evidence of bad faith, declining to sanction Cohen or end his court-supervised release early.
Google DeepMind's AI Football Coach Passes Expert Evaluation
Google DeepMind developed TacticAI, an AI system that provides suggestions for player positioning during football corner kicks, in collaboration with Liverpool FC over three years.
TacticAI analysed 7,176 corner kick situations from the English Premier League between 2020 and 2023, and its suggestions were favoured 90% of the time by 5 experts (3 data scientists, 1 video analyst, 1 coaching assistant) at Liverpool FC over existing strategies.
AI Pioneers Novel Pathways in Antibody Development
Scientists from the University of Washington, employing the RFdiffusion AI tool, showcased the creation of 100 unique antibody designs targeting proteins from pathogens like SARS-CoV-2, as detailed in a bioRxiv1 preprint.
Challenges such as enhancing binding affinity and adapting designs for human use remain, particularly for complex targets like G-protein coupled receptors.
Google's AI Flood Forecasting Expands Globally
Google's AI accurately predicts floods in 80 countries, offering 7-day warnings and improving safety for 460 million people in regions with limited data.
Google's crisis response benefits from partnerships with academics and organisations, like the WMO's Early Warnings for All, which aims to create worldwide climate hazard alerts by 2027.
AI Predicts Post-Treatment Risks for Breast Cancer Patients
AI tools were developed to forecast side effects in breast cancer patients after surgery and radiotherapy; clinical trials will start in France, the Netherlands, and the UK.
Trained on 6,361 patients, the tool predicts arm swelling with 73.4% accuracy, guiding preventive measures and treatment choices.
Stability AI Faces Key Researcher Departures Amid Challenges
Stability AI, the developer of Stable Diffusion, saw lawsuits and major resignations, including Robin Rombach and two fellow researchers, after the company struggled to post a $100 million funding round in 2022.
Garlic Chicken Without Garlic, Amazon Cookbook Controversy
A cookbook from Amazon, "The Complete Crockpot Cookbook for Beginners, 2024 edition," shows signs of being AI-generated, with a fake author photo and a biography with limited online presence.
Negative feedback points out specific issues: advertised recipes are missing, garlic chicken lacks garlic, recipes use rare ingredients, and despite beef images on the cover, there's no beef section.
Astera Labs and Reddit IPOs Highlight Investor Enthusiasm for AI
Astera Labs' IPO saw a 72% surge in share price, closing at $62.03, underscoring the strong market interest in AI-driven companies.
With Astera's valuation reaching around $8.9 billion and Reddit poised to debut, these IPOs may boost confidence among other tech firms considering going public in the AI sector.
Love the AIML newsletter? Please share it with your friends, family, and colleagues who want to stay on top of AI. I hope you have a great day. Follow us.