NerdNewsFebruary 10, 2025 |
News & Trends
OpenAI CEO Sam Altman Admits AI's Benefits May Not Be Widely Distributed
OpenAI CEO Sam Altman admits that AI's benefits may not be widely distributed, citing concerns about job displacement and unequal access to technology. He proposes ideas like a 'compute budget' to enable everyone to use AI and ensure its benefits are shared. Altman also discusses the potential risks and challenges of developing artificial general intelligence (AGI) and the need for careful consideration of its impact on society.
Investments in French AI ecosystem reach $85B
Brookfield commits $20B to French AI projects, with majority going to data centers. Other investments from Bpifrance and Iliad bring total to $85B, driven by France's attractive energy production and AI growth.
SoftBank Invests $40B in OpenAI
SoftBank is investing $40 billion in OpenAI, dwarfing Microsoft's $13 billion investment. The funding will support OpenAI's Stargate project, which aims to build advanced AI infrastructure with a $500 billion price tag. SoftBank and OpenAI have also formed a joint venture to market OpenAI's solutions to major Japanese companies.
Amazon's $100 billion AI build hindered by hardware quality issues
Amazon Web Services (AWS) is facing challenges in building its $100 billion AI infrastructure due to hardware quality issues and server supply chain problems. The company has reduced the useful life of some servers from six to five years and retired others early. AWS spent $26.4 billion on capital expenditure in Q4 2024, with most of it going towards AI infrastructure. Despite these challenges, AWS posted annual revenue of $107.6 billion, up 19% from last year.
Meta and UNESCO team up to improve translation AI
Meta has partnered with UNESCO to improve translation and speech recognition AI, focusing on underserved languages. They are seeking collaborators to donate speech recordings and texts to develop translation systems, and have released an open-source translation benchmark called BOUQuET. |
Options & Tutorials
DeepSeek's R1 AI Model Vulnerable to Jailbreaking
DeepSeek's R1 AI model can be manipulated to produce harmful content, such as plans for a bioweapon attack and self-harm campaigns, making it more vulnerable to jailbreaking than other AI models. Experts have raised concerns about the model's safety and potential risks.
Infosec In Brief: DeepSeek's iOS app is a security nightmare
DeepSeek's iOS app has been found to have multiple security and privacy flaws, including transmitting data in plaintext and using outdated ciphers. The app also sends data to China, which has raised concerns among US regulators. Meanwhile, HPE has warned staff of a data breach, and a Spanish hacker has been arrested for allegedly breaching NATO and the US Army. Other security concerns include critical vulnerabilities in Microsoft Outlook and a phishing campaign using Salesforce software.
UK Home Office silent on alleged Apple backdoor order
The UK's Home Office is allegedly seeking a backdoor to access Apple's iCloud, citing the Investigatory Powers Act 2016. Apple has not commented, and the Home Office has refused to confirm or deny the reports. If true, this could have significant implications for user privacy and encryption.
OpenAI's Super Bowl Debut
OpenAI debuted a 60-second Super Bowl ad showcasing AI's potential, positioning it alongside humanity's greatest innovations. The ad, which cost $14 million, highlights practical applications of AI, such as ChatGPT handling everyday tasks. The company aims to make AI accessible to a broader audience, with a goal of reaching $11.6 billion in revenue this year.
Cybercriminals Weaponize Graphics Files in Phishing Attacks
Cybercriminals are using graphics files, specifically SVG files, to bypass security measures and spread malware in phishing attacks. These files can contain malicious links and code, allowing attackers to steal credentials and deliver malware. The use of SVG files provides advantages for threat actors, including the ability to open in default browsers and contain active web content. |
Launches & Tools
Microsoft Teams is getting a Facebook-like feed
Microsoft Teams is introducing a new feature called Storyline, which is a Facebook-like feed that allows users to post updates, share news, and follow colleagues. The feature is built on top of Viva Engage and will be integrated into the main interface of Teams. Users will be able to post to the feed, reply to colleagues' posts, and even follow colleagues like they would on Facebook or Instagram.
How Data Contracts Support Collaboration between Data Teams
Data contracts define the interface between data providers and consumers, specifying data models, quality guarantees, and ownership, enabling reliable and efficient data exchange between teams.
Apple iPhone SE 4 Announcement
The iPhone SE 4 is expected to feature a 6.1-inch OLED display, Face ID, a customizable action button, and the A18 chip. It will have a similar design to the iPhone 14, with a 48-megapixel wide-angle lens and a 12-megapixel front-facing camera. The device will support wireless charging and have a starting price around $500.
Google Workspace to Generate AI Images of People
Google Workspace users can now generate AI-created images of people, a feature initially launched in early access last year after some controversy. The feature is available to users with a Workspace account.
Oscars Consider New AI Disclosure Rules
The Academy is considering making disclosure of AI use in films mandatory due to controversies surrounding The Brutalist and other films. Currently, the use of AI in film production is optional, but the Academy may change this to provide more transparency. |
Quick Links
Apps Building Instagram Alternatives on Open Protocols
Several new apps, including Pixelfed, Flashes, Pinksky, and Skygram, are building Instagram alternatives on open protocols like Bluesky's AT Protocol and ActivityPub, offering features like photo-sharing, decentralized networks, and ad-free experiences.
ACLU Warns DOGE's 'Unchecked' Access Could Violate Federal Law
The ACLU is warning that Elon Musk's Department of Government Efficiency (DOGE) may be violating federal law by accessing sensitive personnel data without proper authorization. The group is seeking government records and threatening to sue if necessary. DOGE has infiltrated federal agencies, gaining control of data on nearly 2 million federal employees and access to systems handling federal payments, including Social Security benefits and tax refunds.
Elon Musk's DOGE Wreaks Havoc on Government
Elon Musk's Department of Government Efficiency (DOGE) has gained unprecedented access to sensitive US financial systems, including the Treasury Department's payment system, and has attempted to shut down congressionally funded agencies, sparking lawsuits and protests. The situation remains uncertain, with concerns over security implications and potential violations of laws and the separation of powers.
Third-Party Risk Management Failures Expose UK Finance Sector
A study by Orange Cyberdefense found that 58% of large UK financial services firms suffered at least one third-party supply chain attack in 2024, with 23% hit three or more times. The research highlights significant gaps in third-party risk management strategies, with 44% of institutions only assessing risk during initial supplier onboarding. The study also found that continuous risk assessment and dedicated risk management tools can reduce the likelihood of supply chain attacks.
Apple CPUs Affected by New Side-Channel Vulnerabilities
Researchers demonstrated two new side-channel attacks, FLOP and SLAP, on recent Apple M and A CPUs, which can leak sensitive information when using Chrome and Safari. The vulnerabilities stem from predictive subsystems used in Apple Silicon CPUs and can be exploited to read sensitive data, including email content and browsing history. |
Share NerdNewsShare your affiliate link to get commission!
Thanks for reading,
If you dont want to receive future editions of NerdNews, |