The afternoon briefing.
A company reportedly blew $500M on Claude due to uncapped usage, Microsoft teases new Surface hardware, and a 17 million device botnet is dismantled.
AI Cost Management. A company reportedly spent $500 million on Claude in one month due to uncapped AI usage, highlighting significant cost management challenges. This incident underscores how AI productivity promises can quickly turn into runaway expenses without proper expertise. Companies face a trade-off between investing in AI tokens and human resources, as AI costs often exceed initial expectations. Wix also cited rapid AI evolution for workforce reductions.
Model Development. OpenAI continues to refine its offerings, updating GPT-5.5 Instant for improved readability and phasing out older models. For public good, OpenAI is making its life sciences AI model, GPT-Rosalind, freely available for biodefense and pandemic preparedness. Liquid AI has also unveiled its new 8B-A1B MoE model, trained on 38 trillion parameters, pushing large language model development. Many employees, however, still lack formal training for AI tools.
Policy and Hardware. Ethical implications of AI are growing, with the UK planning to use AI to estimate asylum seekers' ages next year. Regulatory scrutiny drives innovation, as NVIDIA introduces a toolkit to automate AI model documentation. In hardware, Acer is launching a Linux handheld for streaming PC games. AI chip startup Groq is reportedly raising $650 million for AI inference, and Microsoft teases new Surface hardware.
Digital Security. The digital security landscape remains active, with law enforcement successfully dismantling a botnet comprising over 17 million compromised devices, reportedly linked to a Russia-based residential proxy network. Meanwhile, genetics company 23andMe is facing a lawsuit over a "disturbing" DNA data breach from 2023, with the California AG alleging the company downplayed the mega-leak. These events highlight ongoing threats from large-scale cybercrime and the critical importance of data protection and accountability.
Company reportedly spent $500 million on Claude after failing to cap AI usage
An unnamed company allegedly blew half a billion dollars on Claude licenses in a single month because nobody set usage limits. This case shows that without real AI expertise, productivity promises can turn into runaway costs.
Tokens or humans? The new corporate trade-off as AI costs rise
AI is costing far more than companies expected, forcing CFOs into a new trade-off between AI tokens and human employees. This poses a risk the market has not yet priced in.
Wix CEO cites 'fast evolution of AI capabilities' in 20% workforce cut announcement
Wix announced a 20% reduction in its headcount, with the CEO citing the rapid evolution of AI capabilities as a contributing factor. Dollar-Shekel currency issues were also mentioned in the decision.
Employees left to navigate AI tools on their own as businesses fail to implement proper training
Only about half of workers regularly using AI have received formal training from their employers, leading to a 'game of chance' in AI adoption. Businesses are failing to implement proper training for their employees.
OpenAI gives GPT-5.5 Instant a readability upgrade while phasing out two older models
OpenAI is updating GPT-5.5 Instant for more natural responses and dropping the Canvas feature from its latest models. The company is also retiring the older o3 and GPT-4.5 models from ChatGPT by August 2026.
OpenAI offers life sciences AI model for free to help governments prepare for pandemics
OpenAI is offering its life sciences model, GPT-Rosalind, for free through the new Rosalind Biodefense program, aimed at pandemic preparedness. Early partners include Lawrence Livermore National Laboratory and Johns Hopkins.
Liquid AI reveals 8B-A1B MoE model trained on 38T parameters
Liquid AI has announced its new 8B-A1B Mixture-of-Experts (MoE) model, which was trained on a substantial 38 trillion parameters. This development marks a significant step in large language model research.
AI will be used to estimate age of asylum seekers from next year in the UK
The UK government plans to implement AI technology next year to estimate the age of asylum seekers. This policy aims to address challenges in age verification for individuals seeking refuge.
NVIDIA MCG Toolkit automates AI model documentation for regulatory compliance
NVIDIA has introduced the MCG Toolkit to automate AI model documentation, addressing increasing regulatory scrutiny under frameworks like California's AB-2013 and the EU AI Act. This tool helps software teams produce comprehensive, auditable model documentation.
Acer launches a Linux handheld for streaming your PC games
Acer is launching the Nitro Blaze Link, a Linux-based handheld device designed primarily for streaming PC games. It features a 7-inch display and Wi-Fi 6, positioning it as a companion device for PC gaming.
AI chip startup Groq reportedly raising $650M to focus on AI inference
AI chipmaker Groq is reportedly seeking to raise $650 million in internal funding as it shifts its focus from hardware to AI inference. This move follows a significant period of growth in the AI sector.
Microsoft teases new Surface hardware and ‘a new era of PC’
Microsoft's Windows and Surface chief Pavan Davuluri has teased new Surface hardware, hinting at a "new era of PC" for developers. The announcement is expected at Microsoft's Build developer conference next week.
Botnet of more than 17 million devices dismantled
A botnet comprising over 17 million compromised devices has been successfully dismantled by law enforcement. The network was reportedly tied to a Russia-based residential proxy service.
23andMe inherits lawsuit over 'disturbing' DNA data breach
Genetics company 23andMe is facing a lawsuit from the California Attorney General over a 2023 DNA data breach. The lawsuit alleges the company downplayed the mega-leak and paid a ransom to the attacker.