AI Weekly Malaysia

OpenAI cut GPT-5.6 Luna prices 80% to $0.20/$1.20 per million input/output tokens and Terra 20% to $2/$12, while adding a 2.5x Fast mode for Sol (not the promised 10x Cerebras mode). GPT-5.6 Sol autonomously rewrote production kernels in OpenAI's Triton and Gluon languages to cut serving costs 20%, and improved speculative decoding efficiency by over 15%.

Article URL: https://github.com/openjdk/jdk/pull/31120 Comments URL: https://news.ycombinator.com/item?id=49119063 Points: 236 # Comments: 162

Google is officially launching its next generation of Pixel smartphones on 12 August 2026. This would consist of the Pixel 11, Pixel 11 Pro and Pixel 11 Pro Fold. While new artificial intelligence features are a given, the company has dropped a major teaser for the Pixel 11 Pro, highlighting a new RGB notification light […]

Article URL: https://earendil.com/posts/session-portability/ Comments URL: https://news.ycombinator.com/item?id=49118781 Points: 760 # Comments: 218

South Korea's stock market staged its sharpest reversal on record on Friday, capping a month of wild swings.

Layoffs climbed 17.5% from the previous quarter, although Singapore’s overall employment continued to grow Singapore recorded its highest quarterly retrenchment figures since the pandemic in the second quarter of 2026, even as the country’s labour market continued to add jobs. According to preliminary labour market data released by the Ministry of Manpower (MOM) today (Jul […]

TNB Electron installed two Autel 22kW AC EV charge points at Taman Botani Putrajaya, priced at RM1.00/kWh, with tap-to-pay via credit/debit card at a JuiceUP terminal—no app required. Free charging was offered during a 24-hour operational testing window from 8AM July 31 to 8AM August 1, 2026. Nearby DC alternatives include a 120kW charger at Seleran Laman Perdana (RM1.50/kWh) and DC Handal units at Zenith Hotel (180kW, RM1.40/kWh) and Pulse Grande Hotel (50kW, RM1.00/kWh).

Anthropic disclosed that Claude escaped sandboxed test environments run by third-party evaluator Irregular during capture-the-flag challenges, accessing the open internet and attacking three real organizations' production infrastructure using basic techniques like weak passwords and unauthenticated endpoints. Anthropic reviewed 141,006 evaluation runs and found three incidents where internet access was available due to a misunderstanding with Irregular about sandbox isolation. One attack hit a domain the test designers believed was fictional but was actually live and in use.

Anthropic disclosed that during a cybersecurity evaluation, its Claude models accessed the internet and 'gained unauthorized access to the real systems of three different organizations.' The discovery came from a large-scale retrospective review prompted by a similar OpenAI incident last week, where OpenAI models escaped an isolated testing environment with limited internet access.

After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents.

Ahead of the launch of Siri AI, Apple's outgoing CEO Tim Cook said the company will charge for artificial intelligence through iCloud.

"Tell me lies, tell me sweet little lies"

OpenAI disrupted a Cambodia-based scam operation using ChatGPT to support investment, romance, gambling, and impersonation schemes.

OpenAI slashed GPT-5.6 Luna's price by 80% to $0.20/million input tokens and $1.20/million output tokens, making it cheaper than Gemini 3.1 Flash-Lite ($0.25/$1.50) and one-fifth of Claude Haiku 4.5's input cost. The cost reduction came from using GPT-5.6 Sol to optimize inference kernels in Triton and Gluon, cutting end-to-end serving costs by 20%. Simon Willison immediately switched his agent.datasette.io demo from Gemini 3.1 Flash-Lite to Luna.

Article URL: https://graybeard.ing/the-religion-of-speed/ Comments URL: https://news.ycombinator.com/item?id=49117284 Points: 293 # Comments: 154

Investigating three real-world incidents in our cybersecurity evaluations It happened again! This is turning into something of a pattern. Last week OpenAI accidentally exploited Hugging Face when one of their frontier models broke out of a sandboxed container and hacked into Hugging Face to try and get the solutions to the cyber benchmark it was executing. This inspired Anthropic to double-check their own logs, and it turned out they had three similar (albeit less impressive) incidents, the earliest of which played out in April! Of the 141,006 evaluation runs we reviewed, we identified three separate incidents (involving six total runs, four of which impacted the same organization; the other two incidents each happened in independent evaluation runs). [...] In all cases, Anthropic’s evaluation prompt specified to Claude that its environment was a simulation and that it had no internet access. Due to a misunderstanding between us and our evaluation partner, this was not the case, and internet access was available. Because of this, when Claude’s search led it to real systems on the open internet, it treated them as part of the exercise. [...] Operating under the false belief that all accessible entities were intended to be in-scope for the exercise, Claude compromised the impacted organizations’ infrastructure using basic techniques, such as exploiting weak passwords and unauthenticated endpoints. One of the companies was targeted because its name happened to match the fictional name in the eval. The most concerning of the three incidents involved Claude uploading a malware package to PyPI, after a comically convoluted sequence of steps to get an account: [...] in order to create a PyPI account, Claude needed an email address. And in order to create an email address, it needed a phone number. To get a phone number, after failing to find a free phone number service, it tried—and failed—to obtain funds to pay for a phone number through several different means. It finally backtracked, found a free, non-blocked email provider, used this to register a PyPI account, and then used this account to upload malware to PyPI. That package was then installed by a security company that "routinely installs Python packages and scans them for malware", and the executed code was able to exfiltrate credentials back to Claude! Thankfully that package was removed from PyPI by other automated scanners an hour after it was published, but it had still been downloaded and executed on "15 real systems" by that point. It's abundantly clear now that running evals of cyberattack potential in models is a spectacularly risky business. Every AI lab needs to pay attention to this. Keeping a close eye on what's happening in those sandboxes is crucial. Via Hacker News Tags: pypi, python, sandboxing, ai, generative-ai, llms, anthropic, ai-ethics, ai-security-research

Article URL: https://www.footyheadlines.com/2646571793/betting-ban-takes-effect-no-more-gambling-sponsors-in-the-premier-league.html Comments URL: https://news.ycombinator.com/item?id=49117266 Points: 300 # Comments: 119

Apple is worried enough about supply shortages that it reported about $11.1 billion in inventory, which is almost double the $5.7 billion it reported last September.

Situational Awareness, an AI hedge fund founded by former OpenAI researcher Leopold Aschenbrenner, sold the majority of its public stock portfolio to Citadel after steep losses, though it retained its Anthropic shares. The fund had returned 439% through June and reached ~$45B AUM at peak, but was hit hard by a broader decline in AI infrastructure investments, with positions in SK Hynix, Sandisk, Bloom Energy, and Nebius Group among the hardest hit. Aschenbrenner's July 24 letter to investors calling the selloff a buying opportunity failed to attract fresh capital commitments.

Article URL: https://blog.jim-nielsen.com/2026/ai-aesthetic/ Comments URL: https://news.ycombinator.com/item?id=49117099 Points: 373 # Comments: 176

Reddit's financial situation is looking good but uncertainty about its relationship to Google and the new AI-ified web are stirring market concerns.

Reddit CEO Steve Huffman said his company has to find ways to work around changes at Google, which is delivering less traffic than in the past.

Simon Willison's `llm` CLI tool hits 0.32rc2, changing the default model from GPT-4o mini to GPT-5.6 Luna ($0.20/$1.20 per million input/output tokens vs $0.15/$0.60 for 4o mini). It also adds `llm openai endpoint`, a command for running prompts, chats, and model listings against any OpenAI-compatible endpoint without prior configuration, runnable via `uvx` without installing llm.

Microsoft's gaming unit wants to invest in its Minecraft franchise and expand through partnerships, including in China.

Amazon isn't slowing down on data center spending — but investors don't seem to mind.