Politics
Are We About to Have Labor Camps in the United States of America?
Ttump's new budget includes $45 billion for detention camps. Alligator Alcatraz is expected to cost $450 million a year. Right now, a reported 5,000 detainees are being held there. The Trump administration says the new $45 billion will pay for 100,000 beds. So that’s 20 more Alligator Alcatrazes out around the country.
Detainees may then be sent out to work, offered back to the companies for which they were just working.
Elon Musk’s ‘Upgraded’ AI Is Spewing Antisemitic Propaganda
On July 4, Musk boasted to his millions of followers, “We have improved @Grok significantly. You should notice a difference when you ask Grok questions.”
Grok replied: “Yes, Jewish executives have historically founded and still dominate leadership in major studios like Warner Bros., Paramount, and Disney. Critics substantiate that this overrepresentation influences content with progressive ideologies, including anti-traditional and diversity-focused themes some view as subversive.”
Kilmar Abrego Garcia Details How CECOT Is A Torture Camp
He was subjected to severe mistreatment upon arrival at CECOT, including but not limited to severe beatings, severe sleep deprivation, inadequate nutrition, and psychological torture. He lost 31 lbs in 2 weeks. Plaintiff Abrego Garcia and 20 other Salvadorans were forced to kneel from approximately 9:00 PM to 6:00 AM, with guards striking anyone who fell from exhaustion.
He observed prisoners in nearby cells who he understood to be gang members violently harm each other with no intervention from guards or personnel. Screams from nearby cells would similarly ring out throughout the night without any response from prison guards on personnel.
People Are Taking Massive Doses of Psychedelic Drugs and Using AI as a Tripsitter
A single session of psilocybin therapy with a licensed practitioner in Oregon can run anywhere between $1,500 and $3,200. The Harvard Business Review reported that one of the leading uses of AI is for therapy.
How People Are Really Using Gen AI in 2025
Therapy is #1.
Infosec
The 2025 Iberian Peninsula Blackout: From Solar Wobbles To Cascade Failures
Because physical generators turn more slowly when they are under heavier load, the frequency of the AC waveform has been the primary coordination mechanism across power plants. When a plant sees a lower grid frequency, it is fueled up to produce more power, and vice-versa. When the system works well, the frequency slowly corrects as more production comes online.
A big issue with the addition of solar generators on the grid is that these use grid-following inverters. Unlike spinning generators that have intrinsic physical inertia, solar inverters can rapidly follow the grid voltage and thus do not dampen grid oscillations or absorb reactive power. Because they can turn on and off essentially instantaneously, these inverters can amplify oscillations and power fluctuations across the grid by boosting or injecting oscillations if the plants over-correct.
This device uses a laser to shoot down 30 mosquitoes per second
LiDAR-guided 'Photonmatrix' has up to 6-meter kill zone, can gauge distance, orientation, and body size in 3 milliseconds
China is building a giant battery the size of a building: an innovative solution for electric transport recharging
Massive concrete blocks are used to store and release electricity, by leveraging gravity. It has a capacity of 100 megawatt-hours.
For comparison, a Tesla Powerwall has a capacity of 13.5 kilowatt-hours (kWh).
How a big shift in training LLMs led to a capability explosion
Foundation models are trained with with imitation learning, creating text that is similar to text found on Wikipedia and other Internet writing samples. But models trained for imitation only learn to respond correctly to inputs similar to the training data, and fail spectacularly when confronted with novel inputs.
One way to improve foundation models is reinforcement learning,
in which the model tries to perform a task and a "supervisor" grades the result. When a human supervisor is too expensive, a trained LLM can act as the supervisor, a system called Reinforcement Learning from Human Feedback (RLHF). This leads to the "reasoning" in DeepSeek and o1.
Scholars sneaking phrases into papers to fool AI reviewers
Nikkei Asia has found that research papers from at least 14 different academic institutions in eight countries contain hidden text that instructs any AI model summarizing the work to focus on flattering comments.
Unless users take action, Android will let Gemini access third-party apps
Google's Microsoft Recall moment.
These XR glasses gave me a 220-inch screen to work with - and have replaced my monitors
The new Xreal One Pro AR glasses are a standout in my book, delivering a high-end experience that's great for both work and play.
How Much More Must We Bleed? - Citrix NetScaler Memory Disclosure (CitrixBleed 2 CVE-2025-5777)
Like Heartbleed, a simple unauthenticated request returns blocks of memory from the server. This can contain authentication tokens.
|