Good morning! This is your daily ☕️ Techpresso.
On this day, 59 years ago, Electronics magazine published an article by Gordon Moore, who predicted that transistor density on integrated circuits would double every eighteen months for at least ten years, a theory now known as Moore's Law, which has largely remained accurate.
In today's Techpresso:
💰 Google mistakenly erased $135 billion from customer account
📱 Apple planning AI-driven overhaul for Siri to manage individual apps
🇺🇸 TikTok to create a US-only algorithm
🦷 World's first tooth-regrowing drug approved for human trials
📚 Perplexity launches AI publishing platform
🎓 OpenAI launches ChatGPT Edu for universities
🎁 + 8 other news you might like
🔮 + 3 handpicked research papers and tools
💰 Google mistakenly erased $135 billion from customer accountLINK
In May, a Google Cloud mishap led to the deletion of Australian pension fund UniSuper's account and backups, freezing operations for two weeks and affecting $135 billion in assets.
Google explained the issue occurred due to an employee failing to fill in a parameter, causing UniSuper's private cloud to be automatically deleted after a fixed term.
Google has now fixed the problem, removed the faulty tool, and assured robust safeguards are in place to prevent similar incidents from affecting other customers.
📱 Apple planning AI-driven overhaul for Siri to manage individual appsLINK
Apple is planning a major AI enhancement for Siri, allowing it to manage specific functions within iPhone apps using voice commands, expected to be announced with iOS 18.
At first, the upgraded Siri will support only Apple-developed apps with "hundreds" of commands, with plans for handling multiple commands in a single request in the future.
Apple's broader AI strategy includes potential integration of OpenAI's technologies and ensuring data security with M2 Ultra chips, as part of an announcement at the WWDC on June 10.
🇺🇸 TikTok to create a US-only algorithmLINK
According to Reuters, TikTok has been working since last year to create a US-specific recommendation algorithm, separate from Douyin’s, in an effort to show its independence from ByteDance in China.
Unnamed sources suggest the project may take more than a year and could reduce TikTok’s reliance on ByteDance’s engineering resources, although TikTok has denied these claims, calling the report misleading and factually inaccurate.
TikTok has previously attempted to assure lawmakers of its US independence through initiatives like "Project Texas," yet skepticism remains regarding the company's ability to avoid a potential ban by January 2025.
🦷 World's first tooth-regrowing drug approved for human trialsLINK
The world's first drug that can regrow teeth has been approved for human trials and was developed by Japanese researchers led by Dr. Katsu Takahashi.
The drug works by deactivating the USAG-1 protein, which normally inhibits tooth growth, leading to new teeth growing; it has been successful in animals with no serious side effects.
Clinical trials will start in September and will include adults missing at least one molar, children with congenital tooth deficiencies, and older adults missing one to five permanent teeth due to environmental factors.
📚 Perplexity launches AI publishing platformLINK
Perplexity introduced 'Pages', an AI-driven feature that creates customizable webpages for research and writing based on user prompts.
The 'Pages' tool gathers information using Perplexity's AI models, organizes it into sections with citations, and tailors content for different audience levels but requires new prompts to correct errors.
The Verge tested 'Pages' on a complex topic and noted its difficulty with intricate subjects, showing it might struggle with in-depth research despite effectively explaining fundamental concepts.
🎓 OpenAI launches ChatGPT Edu for universitiesLINK
OpenAI has introduced ChatGPT Edu, a new version of ChatGPT, aimed at improving the academic and operational efficiency of universities with advanced text interpretation and data analysis capabilities, and support for over 50 languages.
This version was created following the success of ChatGPT Enterprise in academic institutions such as Oxford, Wharton, and ASU, where it has been used for tasks ranging from large dataset analysis to personalized language practice.
ChatGPT Edu provides affordable access to the GPT-4o model, enhanced message limits, robust security measures, and assures that data and conversations will not be used for training OpenAI models, ensuring a safe environment for educational use.
Other news you might like
OpenAI says it stopped multiple covert influence operations that abused its AI models.LINK
Google says it fixed the AI Overviews everyone has been roasting for weeks.LINK
Anthropic’s Claude AI now autonomously interacts with external data and tools.LINK
Misinformation works, and a handful of social ‘supersharers’ sent 80% of it in 2020.LINK
Google and Magic Leap sign deal for AR experiences.LINK
Spotify offers Car Thing refunds as it faces lawsuit over bricking the streaming device.LINK
Basalt plans to ‘hack’ a defunct satellite to install its space-specific OS.LINK
AI-directed drones could help find lost hikers faster.LINK
Latest research and tools
Wadzilla: converts DOOM WAD files into ZIL text output, allowing players to experience DOOM through the text-based interface of an Infocom Z-Machine, similar to playing the classic game Zork.LINK
NPGA: Neural Parametric Gaussian Avatars - high-fidelity digital faces are created using advanced neural network techniques.LINK
NeVerMore: Exploiting RDMA Mistakes in NVMe-oF Storage Applications: the study exposes vulnerabilities in RDMA implementations within NVMe-over-Fabrics storage systems, leading to security concerns.LINK
Want to get the latest news differently? Find us on:
See you tomorrow for a new dose of ☕️ Techpresso!
-min.png){kind=small}
.png){kind=small}
