29 July 2025
Topic: Companies And Business
Acree opens up new enterprise-focused, customizable AI model AFM-4.5B trained on ‘clean, rigorously filtered data’
Geared toward Acree's growing list of enterprise customers and their needs and wants — specifically, a model trained without violating IP....
TL;DR
Arcee.ai opens up its AFM-4.5B model for limited free usage by small companies, providing a cost-efficient, customizable, and compliant alternative to mainstream large language models.
Key Takeaways:
Key Takeaways:
- AFM-4.5B is a 4.5-billion-parameter model designed for real-world enterprise use, combining cost efficiency, regulatory compliance, and strong performance in a compact footprint.
- The model is available for deployment across various environments, including cloud, smartphones, and edge hardware, and is geared towards Arcee's growing list of enterprise customers and their needs.
- Arcee.ai aims to carve out a new category of enterprise-ready language models with AFM-4.5B, which meets enterprise needs for speed, sovereignty, and scale through competitive benchmarks, multilingual support, strong compliance standards, and flexible deployment options.
Anthropic reportedly nears $170B valuation with potential $5B round
The new round is being led by Iconiq Capital, with the possibility of a co-lead investor joining....
Microsoft in talks to maintain access to OpenAI’s tech beyond AGI milestone
Microsoft is reportedly in advanced talks with OpenAI for a new agreement that would give it ongoing access to the startup’s technology even if OpenAI...
TL;DR
Microsoft is reportedly in advanced talks with OpenAI for a new agreement that would grant it ongoing access to OpenAI's technology, potentially clearing a key hurdle in OpenAI's transition to a fully commercial enterprise.
Key Takeaways:
Key Takeaways:
- Microsoft seeks a bigger stake in OpenAI's restructured company and wants to secure its ongoing access to OpenAI's tech beyond the current deal.
- OpenAI aims to shake off its constraints as a mission-driven nonprofit by transitioning to a for-profit company and guaranteeing Microsoft deploys OpenAI's technology safely.
- Bloomberg reports that Microsoft is seeking an equity stake in OpenAI in the low- to mid-30% range.
Sparrow raises $35M Series B to automate the employee leave management nightmare
Sparrow raises $35M Series B to scale AI-powered employee leave management platform that has grown 14x since 2021, serving 1,000+ companies and saving...
TL;DR
Sparrow, an employee leave management technology company, raises $35 million in Series B funding to aid in navigating complex state and local leave regulations.
Key Takeaways:
Key Takeaways:
- The company has grown revenue 14x since its Series A, now serving over 1,000 customers and managing leave for over 500,000 employees.
- Sparrow's AI-powered platform has saved customers over $200 million in payroll costs by ensuring employees receive proper wage replacement from state agencies and insurance providers.
- The company plans to expand its compliance capabilities internationally, starting with global leave solutions and eventually moving into adjacent compliance areas like accommodations and immigration.
AI vs. AI: Prophet Security raises $30M to replace human analysts with autonomous defenders
Prophet Security raises $30 million to launch a fully autonomous AI cybersecurity platform that investigates and responds to threats without human int...
TL;DR
Prophet Security raises $30 million in Series A funding to accelerate the adoption of autonomous AI systems for cybersecurity defense, aiming to automate manual tasks and reduce resource constraints.
Key Takeaways:
Key Takeaways:
- Organizations receive an average of 960 security alerts daily, with up to 40% going uninvestigated due to resource constraints.
- Prophet's agentic AI SOC Platform has performed over 1 million autonomous investigations, saving 360,000 hours of investigation time and reducing false positives by 96%.
- The company's approach is expected to reshape the cybersecurity landscape, with 40% of large enterprises predicted to deploy agentic AI systems in their SOCs by 2025.
YouTube will identify and restrict minors’ accounts with AI
YouTube will soon begin using AI-powered age estimation technology to figure out which users are under the age of 18. Starting August 13th, YouTube wi...
TL;DR
YouTube will start using AI-powered age estimation to restrict minors' access to age-restricted content, starting August 13th.
Key Takeaways:
Key Takeaways:
- YouTube will automatically apply restrictions to accounts that appear to belong to users under 18, such as blocking age-restricted videos and non-personalized ads.
- This feature will be rolled out to a small set of users in the US first, before being expanded to other countries, and will allow users to verify their age if they think they've been misclassified.
- The change aims to comply with a global crackdown on child safety online, following similar measures introduced in the UK and elsewhere.
Google won’t say if UK secretly demanded a backdoor for user data
Google said it has "never built a backdoor" for its services, but refused to rule out that it had received a secret U.K. surveillance order demanding ...
TL;DR
The U.K. government may be backing down from its demand for a secret backdoor to access customer data worldwide, following a rebuke from the U.S. government.
Key Takeaways:
Key Takeaways:
- U.K. authorities demanded a secret backdoor from Apple to access encrypted customer data, sparking global concerns about user privacy.
- Apple has appealed the legality of the order, and at least one other tech giant, Meta, has said it didn't receive a similar order.
- Google has refused to confirm or deny whether it received a U.K. government order to access encrypted data, citing secrecy obligations.
Big Tech Asked for Looser Clean Water Act Permitting. Trump Wants to Give It to Them
New AI regulations suggested by the White House mirror changes to environmental permitting suggested by Meta and a lobbying group representing firms l...
TL;DR
The Trump administration proposes loosening Clean Water Act permitting processes for data centers, a move critics argue could harm the environment and favor companies like Google and Amazon.
Key Takeaways:
Key Takeaways:
- The proposed changes could allow data centers to skip the current permitting process, which currently requires a case-by-case analysis and can be expensive and time-consuming.
- Critics argue that a nationwide permit for data centers would be inconsistent with the intent of the Clean Water Act and could lead to increased environmental harm.
- The proposal is seen as part of a broader effort by the Trump administration to facilitate the growth of the data center industry, which is driving demand for energy and water resources.
Positron believes it has found the secret to take on Nvidia in AI inference chips — here’s how it could benefit enterprises
The company’s first-generation chips were fabricated in the U.S. using Intel facilities, with final server assembly and integration....
TL;DR
Private chip startup Positron positions itself as a challenger to Nvidia by offering dedicated, energy-efficient, memory-optimized inference chips for large-scale AI deployment.
Key Takeaways:
Key Takeaways:
- Positron's Atlas chip delivers 3.5x better performance per dollar and up to 66% lower power usage than Nvidia's H100.
- The company's next-generation platform, Titan, will support models up to 16 trillion parameters and operate with standard air cooling in conventional data center environments.
- Positron is focusing on memory-first design, betting on the shift from compute-bound convolutional neural networks to memory-bound transformer architectures in AI workloads.
Google’s NotebookLM rolls out Video Overviews
With this new capability, NotebookLM is taking a more visual approach to helping users understand different topics and ideas....
TL;DR
Google rolls out Video Overviews to NotebookLM, enabling users to turn dense multimedia into digestible visual presentations.
Key Takeaways:
Key Takeaways:
- Google is rolling out Video Overviews to NotebookLM, allowing users to generate visual presentations that combine images, diagrams, quotes, and numbers.
- Users can customize their Video Overviews based on specific topics, learning goals, and target audiences.
- Video Overviews are now available to all users in English, and support for more languages will be added soon.
Spotify hints at a more chatty voice AI interface in the future
Spotify says AI will allow for "much more interactive" consumer experiences in its app....
TL;DR
Spotify is exploring the potential of generative AI to create a more conversational interface for its streaming service, leveraging voice data to inform music recommendations and user experiences.
Key Takeaways:
Key Takeaways:
- Spotify's AI DJ feature has generated a new dataset, associating phrases with songs, enabling more interactive experiences.
- The company plans to expand its voice interface capabilities, allowing users to interact with Spotify in plain English.
- Spotify is using generative AI to prototype products and create efficiencies in areas like finance, in addition to enhancing consumer features.
Google’s NotebookLM can now make narrated slideshows with AI
Google’s NotebookLM is getting a new Video Overviews feature that uses AI to create slideshows with narration. The feature is rolling out now in Engli...
TL;DR
Google's NotebookLM is getting a new Video Overviews feature using AI to create narrated slideshows, rolling out now in English with future support for more languages.
Key Takeaways:
Key Takeaways:
- The new Video Overviews feature uses AI to create slideshows with narration and can be used to explain data, demonstrate processes, and make abstract concepts more tangible.
- The Studio tab in NotebookLM is getting a visual refresh and will allow users to create and store multiple studio outputs of the same type in a single notebook.
- The Video Overviews feature includes playback controls like skipping back and forth by 10 seconds and setting playback speed, with additional formats to be introduced in the future.
Google Workspace is rolling out a security update to stop token stealing attacks
Google Workspace is launching a new security measure to help prevent the same type of account takeover attack that impacted Linus Tech Tips. The featu...
TL;DR
Google Workspace is rolling out a security update called Device Bound Session Credentials (DBSC) to prevent token stealing attacks and protect user accounts from account takeover attacks.
Key Takeaways:
Key Takeaways:
- Over 11 million Google Workspace customers can now use passkeys to add an extra layer of security.
- Google Workspace has seen an 'exponential rise' in cookie and authentication token theft over the past couple of years and this trend has intensified in 2025.
- The company recommends enabling passkeys and blocking bad actors from remotely stealing session cookies with the Device Bound Session Credentials feature.
Trump’s AI plan is a massive handout to gas and chemical companies
The Trump administration put out its vision for AI infrastructure in the US last week. It's a dream for the fossil fuel and chemical industries - and ...
TL;DR
The Trump administration is proposing an AI plan that would give massive handouts to gas and chemical companies, and erase bedrock environmental protections in the US, prioritizing fossil fuels over wind and solar energy.
Key Takeaways:
Key Takeaways:
- The plan would accelerate federal permits for data center infrastructure, prioritizing 'dispatchable' power sources like natural gas and coal.
- The Trump administration is also planning to loosen environmental regulations, including modifying rules under the Clean Air Act and Clean Water Act.
- Critics argue that this plan would have devastating environmental consequences, particularly in terms of air and water pollution, and undermine the transition to renewable energy sources.
Anthropic Faces Potentially "Business-Ending" Copyright Lawsuit
Article URL: https://www.obsolete.pub/p/anthropic-faces-potentially-business Comments URL: https://news.ycombinator.com/item?id=44722048 Points: 20 # ...
TL;DR
Anthropic faces potentially 'business-ending' copyright lawsuit for mass use of pirated books, which could exceed a billion-dollar penalty.
Key Takeaways:
Key Takeaways:
- The company is at risk of bankruptcy due to the statutory maximum damages of $150 billion and a likely jury award of $1.5 billion.
- Anthropic's lawyers acknowledge that the number of covered books likely ranges in the millions, further increasing potential damages.
- A precedent-setting loss could expose Anthropic's competitors, including OpenAI and Meta, to similar liability for mass copyright infringement.
Stop promising "unlimited", when you mean "until we change our minds"
Article URL: https://blog.kilocode.ai/p/ai-pricing-playbook-strikes-again Comments URL: https://news.ycombinator.com/item?id=44721003 Points: 70 # Com...
TL;DR
Anthropic adds weekly limits to its high-end AI model, Claude Pro, affecting its heaviest users, in a move some see as part of the 'AI pricing bait-and-switch' playbook.
Key Takeaways:
Key Takeaways:
- Power users and heavy developers are often the first to have their 'unlimited' AI access capped, affecting their workflow and productivity.
- The AI pricing bait-and-switch approach, where promises of 'unlimited' access are later restricted, can damage the trust and growth of the developer community.
- More transparent and predictable pricing models, where users are given control over their usage and costs, can be an alternative to the current industry approach.
TL;DR
NVIDIA announced the new Blackwell B200 GPU, promising significant performance improvements in AI training and inference.
Key Takeaways:
Key Takeaways:
- Claims a 20x performance increase in large language model inference compared to the H100 GPU generation.
- Features a second-generation Transformer Engine and new tensor core technology for enhanced AI capabilities.
- Cloud providers like AWS, Google Cloud, and Azure have already committed to supporting the new Blackwell B200 architecture.
Community talk
AI Tools
Show HN: Xorq – open compute catalog for AI
Hi HN, Hussain and Dan from Xorq here.After years of struggl..
OpensourceShow HN: Terminal-Bench-RL: Training Long-Horizon Terminal Agents with RL
After training calculator agent via RL, I really wanted to g..
OpensourceYouTube Videos
China Just Dropped the Smartest Open Source AI Ever Built (Crushed DeepSeek & Benchmarks)
AI revolutionX
•
Jul 29, 2025
zuckerberg offered a dozen people in mira murati's startup up to a billion dollars, not a single person has taken the offer
Have you noticed Google's AI overviews have gotten dramatically worse recently?
I found out AI detectors mostly flag text as AI if it sounds "too obedient"
Are We on Track to "AI2027"?
I just spoke to an "AI founder" he’s pivoting back to his old services
xAI no longer has researchers.
Has "Think longer" disappeared? Here last night, it's gone for me today.
Claude Code 1.0.62 has a command discovery bug (Windows)