🇬🇧 UK judge warns: Fake AI cases = contempt of court

London’s High Court just fired a warning shot: lawyers citing AI-generated fake cases can now be hit with contempt charges, or even criminal prosecution. Judge Victoria Sharp ripped into lawyers caught using ChatGPT-style tools to fabricate case law, calling it a “serious threat” to justice.

Because apparently, in 2025, hallucinating case law is a thing. And no, "guidance" isn’t going to cut it anymore; regulators are now being told: do something. Next time you walk into court with AI-made nonsense, it might not just be your career on the line. It could be your freedom.

@aipost 🪙 | Our X 🥇

OpenAI slashed o3's price by 80% and dropped a new o3-pro that:

— Thinks longer to boost performance
— Beats rivals on PhD-level math & science
— Can web search and data analysis (but no image generation & canvas)
— Is available to ChatGPT Pro and Team users

@aipost 🪙 | Our X 🥇

Here's how the AI ​​market's traffic share has changed over the past six months:

6 months ago:
ChatGPT: 87.5%
Google: 5.4%
Perplexity: 2.0%
Claude: 1.6%

3 months ago:
ChatGPT: 77.6% (-9.9%)
DeepSeek: 8.1% (new player!)
Google: 4.9%
Grok: 2.7% (comes into play from X)
Perplexity: 2.0%

1 month ago:
ChatGPT: 80.2% (+2.6%)
Google: 6.1% (+1.2%)
DeepSeek: 5.9% (-2.2%)
Grok: 2.4%
Perplexity: 1.6%
Claude: 1.2%

Now:
ChatGPT: 78.9% (-1.3%)
Google: 8.0% (+1.9% )
DeepSeek: 5.3% (-0.6%)
Grok: 2.1% (-0.3%)
Perplexity: 1.7% (+0.1%)
Claude: 1.4% (+0.2%)

Key findings:
- ChatGPT is still the leader, but is gradually losing market share (-8.6% over six months).
- Google is growing steadily (+2.6% since last year).
- DeepSeek had a strong start, but has lost ground recently.
- Grok and Perplexity are holding up, while Claude is showing some growth.

@aipost 🪙 | Our X 🥇

🚥 Google AI overviews faces backlash for dangerous 'hallucinations'

Google’s AI Overviews, designed to give quick answers, has been under fire for spreading false and dangerous advice, like using glue in pizza sauce and citing non existent idioms. The system, powered by Gemini, is causing click-through rates to reputable sources to plummet by 40%–60%.

Despite claims that it broadens user information, data shows hallucination rates as high as 1.8%. Recent testing of OpenAI’s newest models suggests even higher hallucination rates, with up to 48% of responses containing false information.

@aipost 🪙 | Our X 🥇

Kevin Weil says OpenAI prioritizes speed by building specialized AI models rather than packing every capability into one

Each model has its strengths and helps release new features faster, even if it feels confusing at first. Now these capabilities merge into the core model

@aipost 🪙 | Our X 🥇

OpenAI Alexander Embiricos says coding is about to shift completely, with most code soon to be written by agents

These agents won’t just help in your IDE -- they’ll work on their own, embedded in tools without constant prompts. At openAI, agents already generate and merge most

@aipost 🪙 | Our X 🥇

Sam Altman in his blog post: "We're close to building digital superintelligence and already past the event horizon -- the takeoff has begun"

That’s a big statement. Sam seems to suggest they’ve built a self-improving system. Maybe it’s not full AGI yet, but it’s powerful enough to speed up science massively, like doing a year of research in just one month

@aipost 🪙 | Our X 🥇

📱 OpenAI expects subscription revenue to nearly double to $10bn

OpenAI has nearly doubled its annual subscription revenue to $10 billion, driven by strong demand for ChatGPT. Despite the growth, the company is still operating at a loss and relies heavily on investor funding. It aims to reach profitability by 2029, targeting $125 billion in revenue.

OpenAI is expanding through major acquisitions and partnerships, though there are early signs that enterprise AI spending may be slowing.

@aipost 🪙 | Our X 🥇

📱 OpenAI CEO Sam Altman reveals how much water a single ChatGPT question really uses

OpenAI CEO Sam Altman recently revealed that a single ChatGPT query uses only about one fifteenth of a teaspoon of water, equating to roughly 0.000085 gallons. This figure aims to address concerns about the environmental impact of AI, particularly regarding energy and water usage.

Altman noted that each query consumes around 0.34 watt-hours of energy, similar to brief usage of common household appliances. However, the overall water consumption for AI operations can vary significantly due to the cooling needs of data centers. While Altman's statement seeks to ease public worries, it raises questions about the sustainability of AI as it scales.

@aipost 🪙 | Our X 🥇

At WWDC 2025, Apple showed off only a handful of AI upgrades, including:

— New Live translation for FaceTime, Messages, and calls
— Visual intelligence via screenshots
— AI-powered intelligent actions in Shortcuts
— AI "Workout Buddy" on Apple Watch

@aipost 🪙 | Our X 🥇

Google DeepMind and the UK govt dropped "Extract," a Gemini-powered tool to expedite infrastructure and housing decisions

It uses multimodal reasoning to turn complex planning docs – including handwritten notes – into digital data in just 40 seconds

@aipost 🪙 | Our X 🥇

Ilya Sutskever says the real challenge of AI is how fast and extreme its growth will be. It still has limits, but it already speaks, codes, and interacts in ways once impossible

Soon, perhaps within 3 to 10 years, AI will be able to do everything humans can do. The reason: "the human brain is a biological computer, so why can't a digital brain do the same?"

@aipost 🪙 | Our X 🥇

🇪🇺 Europe wants AI that obeys… Nvidia & perplexity just handed them the blueprint

Tired of AI that won’t censor fast enough? Europe is getting its own… thanks to Nvidia and Perplexity. They’re collaborating with French firm H Company to build “sovereign AI” models tailored to comply with strict EU rules. This contrasts sharply with the U.S. tech industry, which is often perceived as the Wild West. Europe blocks significantly more content than the U.S., including political speech and anything that Brussels deems unacceptable, all in the name of “safety” and “truth.”

Instead of struggling to make American models comply, Europe will now have AI pre-trained to adhere to Brussels’s guidelines. This approach is local, open-source, and obedient, precisely what regulators desire. Nvidia benefits from goodwill, Perplexity gains access, and Europe gains control. However, free speech is not included in the roadmap. The focus is on AI that won’t alarm Brussels.

@aipost 🪙 | Our X 🥇

💩 Meta's Zuckerberg hiring for new 'superintelligence' AI team

Meta CEO Mark Zuckerberg is reportedly assembling a new AI team to enhance the company's artificial intelligence capabilities. This initiative aims to drive innovation and improve AI integration across Meta's platforms.

The focus will be on developing advanced AI tools and technologies to bolster user experience and engagement. This move reflects Meta's commitment to staying competitive in the rapidly evolving AI landscape, especially as the demand for AI-driven solutions continues to grow.

@aipost 🪙 | Our X 🥇

🔍 Google offers buyouts to cut headcount as AI spending surges

Google has started offering buyouts to employees across multiple teams including knowledge and information, central engineering, marketing, research and communications. The buyouts, part of efforts to trim headcount after cutting 12,000 jobs in 2023, come as Google ramps up AI infrastructure spending.

Employees have been given the option to voluntarily exit with severance, a preferred strategy as the company shifts resources. Meanwhile, Google demands many remote workers return to the office or risk losing their jobs. Watch for continued changes as Google reorganizes for its AI push.

@aipost 🪙 | Our X 🥇

For the first time, an autonomous drone defeated the top human pilots in an international drone racing competition. This is a novelty.

@aipost 🪙 | Our X 🥇

The meme is more realistic than you might think. So many people use ChatGPT as a therapist.

@aipost 🪙 | Our X 🥇