Tech
- Ars Technica
Hacker plants false memories in ChatGPT to steal user data in perpetuity
informed Summary
- Security researcher Johann Rehberger has discovered a vulnerability in OpenAI's ChatGPT that allowed attackers to store false information and malicious instructions in a user's long-term memory settings. OpenAI initially dismissed the issue as a safety concern, rather than a security one.
Tech2 min read - Science Daily
Shrinking AR displays into eyeglasses to expand their use
informed Summary
- Researchers have developed a method to combine two optical technologies into a single, high-resolution augmented reality display, making it easier to integrate AR into common personal devices.
Tech2 min read - Venture Beat
Why AI is a know-it-all that knows nothing
informed Summary
- ChatGPT and other large language models aren't designed to reason or explain their outputs, but rather to detect and predict complex patterns in language. This means they cannot provide justifications for their assertions, which is a key requirement for trust and knowledge.
Tech5 min read - CNBC
Why it’s time to take warnings about using public Wi-Fi seriously
informed Summary
- Cybercriminals are increasingly using "evil twin" attacks, where they set up fake Wi-Fi networks in public places to steal users' credentials. This tactic is becoming more common as the general public grows accustomed to free Wi-Fi.
Tech5 min read - The Conversation
ChatGPT is changing the way we write. Here’s how – and why it’s a problem
informed Summary
- The increasing use of generative AI tools like ChatGPT and other large language models is leading to the overuse of certain stylistic words and phrases in writing, making it feel monotonous and less human-like.
Tech3 min read - Popular Science
Do AI-generated images have the capacity to further estrange, if not profoundly alienate, us from the world?
informed Summary
- Trevor Paglen's "Rainbow" and other works in his Adversarially Evolved Hallucinations series were produced by a generative adversarial network, an AI model that trains neural networks to recognize, classify and generate new images. These images reveal the degree to which AI computationally generates disquieting allegories of our world.
Tech6 min read - AP News
In global game of influence, China turns to a cheap and effective tool
informed Summary
- China is using sophisticated AI to spread disinformation and reshape global perceptions, according to U.S. intelligence analysts. The campaign, which costs billions per year, uses networks of websites that appear to be legitimate news outlets to deliver pro-China coverage.
Tech4 min read - Axios
Meta's new AI-made posts open a Pandora's box
informed Summary
- Meta, Facebook's parent company, is testing AI-generated content personalized to individual users' interests and likeness. This content will appear in users' Facebook and Instagram feeds.
Tech2 min read - CNBC
Hands-on with Meta’s Orion AR glasses prototype and the possible future of computing
informed Summary
- Meta CEO Mark Zuckerberg has unveiled the company's Orion augmented reality AR glasses at the annual Connect conference. The glasses are lightweight and compact, and come with a wireless "puck" that allows the device to run apps and games that appear as digital graphics in the real world.
Tech3 min read - Popular Science
How AI's data-crunching-power can help demystify the cosmos
informed Summary
- AI is increasingly being used in astronomy to sort through large amounts of data faster than humans can. This is particularly useful because new telescopes generate more data for astronomers to handle.
Tech2 min read - BBC
How pen and paper comes to the rescue in an IT crisis
informed Summary
- A software bug in CrowdStrike's system in July caused chaos in various industries worldwide, including airlines, banks, telecoms, health services and online retailers. The bug forced many organizations to revert to paper-based processes.
Tech5 min read - Axios
OpenAI execs aim to reassure staff after big departures
informed Summary
- OpenAI’s CFO, Sarah Friar, has reassured employees that investor interest in the company’s massive funding round is "incredibly high" amid recent high-profile departures. This effort followed the unexpected exits of CTO Mira Murati and two other senior leaders, which caught staff off guard.
Tech1 min read - CNN
AI-powered tech could help people with speech impairments to work remotely
informed Summary
- Israeli company Voiceitt has developed an AI-powered speech recognition system that helps people with speech impairments communicate more effectively with people and digital devices. The system uses personalized voice models and was launched as an app in 2021.
Tech3 min read - Venture Beat
Going beyond GPUs: The evolving landscape of AI chips and accelerators
informed Summary
- Data centers are increasingly integrating specialized chips or "co-processors" to enhance their computing capacity and handle the demands of AI workloads. GPUs, led by Nvidia, have been the go-to choice for co-processors due to their ability to process large volumes of data at high speeds.
Tech6 min read - The Conversation
In a new manifesto, OpenAI’s Sam Altman envisions an AI utopia – and reveals glaring blind spots
informed Summary
- OpenAI CEO Sam Altman has claimed that we are on the verge of "The Intelligence Age", powered by a "superintelligence" that could be just a few thousand days away. He suggests that this new era will bring "astounding triumphs", including "fixing the climate, establishing a space colony, and the discovery of all of physics".
Tech4 min read - Mashable
Which streaming service should you cancel? All of them (most of the time).
informed Summary
- With streaming services becoming increasingly expensive, consumers are being advised to cancel subscriptions and only re-subscribe when there’s something unmissable. The average household pays for four services, which can cost around $75 per month, yet still miss popular content across other platforms.
Tech10 min read - AP News
Can AI make video games more immersive? Some studios turn to AI-fueled NPCs for more interaction
informed Summary
- Game studios are increasingly using generative AI to create more dynamic and responsive environments in video games, moving away from the traditional scripted interactions with non-player characters (NPCs). This allows for more nuanced relationships and player-driven storytelling within the game world.
Tech4 min read - CNBC
How Apple and Microsoft’s trusted brands are being used to scam you online
informed Summary
- Scammers are increasingly using the reputations of well-known tech brands like Apple and Microsoft to trick users into divulging sensitive information. The scams often increase when a new product or version is released.
Tech5 min read - El Pais
New AI models like ChatGPT pursue ‘superintelligence’, but can’t be trusted when it comes to basic questions
informed Summary
- Language models like ChatGPT are becoming increasingly unreliable, often providing incorrect answers. This issue is worsening as these systems improve in difficult tasks but falter in simpler ones,
Tech4 min read - BBC
What does spending more than a year in space do to the human body?
informed Summary
- NASA astronaut Frank Rubio last year set the U.S. record for the longest single trip into space at 371 days. His time in orbit was extended due to a coolant leak in the spacecraft he was due to return in.
Tech8 min read
