AI security threats, nuclear-powered data centers, and Nvidia's new AI model

Close the AI impact gap with Airtable.

Legacy applications are holding back the pace of your digital transformation.

The future of enterprise operations is AI-powered. Airtable gives you the tools to make that future a reality today. Deploy AI across your organization, automate bespoke processes, and unlock unprecedented efficiency and innovation.

Visit Airtable.com to Learn More

Security threats to AI models give rise to a new crop of startups

As AI adoption grows, security concerns are emerging, particularly around data leakage and adversarial attacks. Startups like Opaque Systems, Credo AI, and Zendata are developing solutions to protect AI systems and sensitive data. Continuous monitoring platforms, such as Protect AI and HiddenLayer, are also stepping in to assess and respond to threats in real time. With companies in regulated industries facing increased risks, the demand for AI security solutions is creating new opportunities for startups. Read more

UK to consult on ‘opt-out’ AI content scraping in blow to publishers

The UK government is set to launch a consultation on an "opt-out" system allowing AI companies to scrape online content unless publishers and creators actively decline. While tech giants like Google support the model, publishers and creatives argue it unfairly burdens them, making it harder to protect their intellectual property. The creative industry prefers an "opt-in" system that would require licensing agreements and compensation. The consultation is expected to begin after the October 30 budget. Read more

EU AI Act checker reveals Big Tech's compliance pitfalls

A new tool testing compliance with the EU AI Act has found that AI models from companies like OpenAI and Meta fall short in areas such as cybersecurity and discriminatory output. Developed by LatticeFlow AI, the "LLM Checker" awarded scores between 0 and 1 across various categories. Some models, like OpenAI’s GPT-3.5 Turbo and Meta’s Llama 2, received low marks for discriminatory output and vulnerability to "prompt hijacking" attacks. With fines of up to 7% of global turnover for non-compliance, tech firms face pressure to fine-tune their models ahead of the AI Act’s full enforcement by 2025. Read more

Nvidia just dropped a new AI model that crushes OpenAI’s GPT-4—no big launch, just big results

Nvidia quietly released Llama-3.1-Nemotron-70B-Instruct, an AI model that surpasses GPT-4 and other industry leaders in key benchmarks. This strategic move marks Nvidia's shift from GPU powerhouse to AI model pioneer, with its new model outperforming competitors in language understanding and generation. Offering businesses a cost-efficient, flexible AI solution, the release intensifies competition in the AI space and signals Nvidia's growing influence beyond hardware. Read more

Will AI ‘compress’ the 21st century?

Top AI leaders are shifting from fear-based narratives to hype, with figures like Sam Altman and Dario Amodei touting AI's potential to revolutionize everything from scientific discovery to governance. Amodei speculates that AI could accelerate progress so dramatically it might "compress" the 21st century, enabling breakthroughs in healthcare, economics, and democracy. While this optimism has its allure, critics warn against assuming AI will reshape humanity without addressing its imperfections. Read more

TSMC bullish on outlook as AI boom boosts Q3 profit past forecasts

TSMC, the world's largest chipmaker, reported a 54% jump in quarterly profit driven by high demand for AI chips. The company, whose clients include Apple and Nvidia, expects 2024 revenue to grow nearly 30%, with AI processors making up a significant share. TSMC’s capital expenditure will rise, with plans for expansion in Arizona and healthy growth forecasted through 2025. This strong performance comes amid industry concerns sparked by competitor ASML’s lower projections. Read more

Google turns to nuclear to power AI data centres

Google has signed a deal with Kairos Power to use small nuclear reactors to meet the growing energy demands of its AI data centres. The first reactor is expected online by 2030, with more planned by 2035. This shift towards nuclear energy aligns with tech companies' efforts to reduce carbon emissions while powering increasingly energy-hungry data centres. The move comes amid a broader industry trend as firms like Microsoft and Amazon also turn to nuclear solutions to power their operations. Read more