Fireworks between Musk and OpenAI

WELCOME

Welcome to the Blacklynx Brief !

We are a slightly shorter read this week and next week - there will be no Blacklynx Brief - unfortunately. It’s been incredibly busy but not so much on the news feed.

We are gearing up for the Game Developers Conference in San Francisco where I’ll be investigating how AI will influence the gaming industry going forward.

Expect a themed edition somewhere towards the beginning of April.

AI NEWS
Musk files lawsuit against OpenAI

• Elon Musk has filed a lawsuit against OpenAI and CEO Sam Altman, alleging that the company has strayed from its original mission to benefit humanity by becoming profit-driven and making GPT-4 proprietary. Musk seeks to redirect OpenAI to its foundational goal of creating open and safe artificial general intelligence (AGI), citing concerns over the board's expertise and the secrecy surrounding the rumored Q* system. The lawsuit also scrutinizes OpenAI's complex structure, which consists of eight sub-companies

• OpenAI has responded to Elon Musk's lawsuit by releasing his own emails in a blog post, which challenge Musk's claims that OpenAI deviated from its mission for profit. The emails reveal Musk's past suggestions for OpenAI to secure substantial funding and his idea for Tesla to acquire the company. The situation escalates as Musk proposes changing OpenAI's name to 'ClosedAI'

  

• OpenAI is gradually introducing a 'Read Aloud' feature for ChatGPT to a small user base without making an official announcement. This feature enables ChatGPT to vocalize responses using AI-generated voices, with users having the option to choose from five different voices: Ember, Breeze, Cove, Juniper, and Sky. The enhancement aims to improve accessibility by allowing ChatGPT to communicate through voice, addressing previous limitations in its interaction capabilities.

• Madonna has incorporated AI-generated video technology from Runway into her Celebration Tour, specifically for the song "La Isla Bonita," replacing previous CGI with more dynamic AI visuals. Runway's platform has also been used by musicians like A$AP Rocky, Kanye West, and Slash, indicating a growing trend in the music industry. This shift highlights a broader acceptance of AI technology as a creative tool among artists, despite initial skepticism regarding its impact on the arts.

• Anthropic has unveiled its Claude 3 model family, introducing variations like Opus, Sonnet, and Haiku that vary in performance, speed, and price. The Opus model notably surpasses GPT-4 and Gemini Ultra in several benchmarks, showing advanced capabilities in expert knowledge, reasoning, coding, and math, along with new vision features comparable to GPT-4V. These models also showcase exceptional context processing with a 200k to 1M context window and have opened immediate API access for developers. This release challenges the current AI leaderboard, positioning Claude 3 prominently in the industry.

  

• The US Army Research Lab is experimenting with commercial AI chatbots like GPT-4 for creating battle strategies in war game simulations such as StarCraft II, with GPT-4 showing better performance among tested models. Despite its advancements, the integration of AI in military planning, supported by collaborations with companies like Palantir and Scale AI, faces ethical, feasibility, and bias concerns from experts. What can go wrong, right ?

• The 2024 Trust Barometer study reveals a significant decline in global trust towards AI companies, with a 26-point gap between the tech industry (76%) and AI (50%). In the U.S., trust in AI has dropped to 35%, a 15-point decrease since 2019, while globally, it fell 8 points to 53%. Main concerns include privacy, dehumanization, societal harm, and insufficient regulation, rather than job loss.

• A Microsoft senior software engineer has reported to the FTC that Copilot Designer, powered by OpenAI's DALL-E, might produce harmful content due to inadequate content filters. Despite the engineer's warnings and the discovery of significant vulnerabilities, Microsoft has not halted the product's distribution. The incident raises questions about Microsoft's dedication to responsible AI, suggesting that safety concerns may be overshadowed by profit motives.

  
  

CYBERSECURITY NEWS
Pegasus Spyware Under Scrutiny

• Brouwerij Moortgat has suffered a cyberattack - halting production across all their plants. They stated that ‘their systems have detected the attack so the system has worked”. But the fact that they are down suggests quite strongly that the system hasn’t “worked”. In the statement it says the IT team is working on a solution. I know from experience that this solution will probably be to go look for the backup tapes only to find they have been encrypted as well.

• A California federal judge ordered NSO Group, an Israeli spyware developer, to submit its secret code in a lawsuit with WhatsApp, marking a significant legal development. This decision could reveal extensive details about NSO Group's Pegasus spyware, previously used to target human rights activists, journalists, and politicians worldwide. However, the court did not compel NSO to disclose its client list or server details, limiting the extent of exposure.

• Russia is accused of releasing an intercepted conversation between German Bundeswehr officials discussing support for Ukraine, aiming to fuel divisions within Germany, especially regarding the supply of Taurus cruise missiles. The incident raises concerns about the security of German military communications, as the conversation was reportedly conducted using Webex, which may lack adequate protection for sensitive discussions. This act is seen as part of Russia's disinformation strategy, potentially altering perceptions and political dynamics in Berlin, while also testing the unity and cyber defense of the targeted parties.

• North Korean hackers targeted South Korean microchip equipment companies, stealing important design and site information, as reported by South Korea's National Intelligence Service. This act of industrial espionage, particularly in the semiconductor sector, could significantly enhance North Korea's technological capabilities and exacerbate global tensions.

• Researchers at JFrog have identified over 100 potentially malicious AI/ML models on the Hugging Face platform, which could install malware, including backdoors, on users' devices. However, the majority of these detections appear to be proof-of-concept, with only 10 models confirmed as truly malicious. This highlights the growing concern of cybersecurity threats within AI/ML model repositories.

• The AlphV ransomware group has reportedly ceased operations and exit-scammed following a substantial ransom payment from Change Healthcare, a US healthcare platform. They posted a deceptive FBI seizure notice on their dark web site and are currently selling the ransomware's source code in hacking forums. This incident came to light after a $22 million ransom payment to the group's Bitcoin wallets was identified by security firms. There are allegations from AlphV members on hacking forums that the group's administrators absconded with the ransom funds without sharing them.

• In an op-ed, Ciaran Martin, former chief executive at the NCSC, as called for the UK government to ban ransomware payments.

• The recent push in Pakistan's Senate to ban popular social media platforms like Facebook, TikTok, Instagram, YouTube, and X (formerly Twitter) spotlights ongoing tensions over internet freedoms in the country. Pakistan employs laws like the Pakistan Electronic Crimes Act (PECA) to penalize online criticism of the government and restrict internet services, reflecting growing control over digital spaces. These actions, including internet shutdowns, particularly affect low-income communities relying on mobile internet, suppress dissent, limit information access, and hinder public participation in critical events such as elections.

BOOK TIP OF THE WEEK
Michael Easter - The Comfort Crisis

This week's book recommendation is Michael Easter's "The Comfort Crisis". The book is basically a trip report on the author's hunting trip up in the north of Alaska. He uses the trip to illustrate the point that our society has become too soft.

Could our sheltered, temperature-controlled, overfed, underchallenged lives actually be the leading cause of many our most urgent physical and mental health issues.

We need challenges in our life. In order to really appreciate food, you need to experience hunger first. In order to appreciate the roof you are sleeping under - you should spend a night out in the biting cold or the pouring rain.

There is truth in that. It builds character but above all infuses you with gratitude.

It's a fascinating investigation into a topic we don't really talk about.

Warmly recommended.

Buy it here.

Goodreads Link

Closing Thoughts

That’s it for us this week.

Can I ask you a favor ?

If you’ve gotten to this part , it means you really read the entire thing.

This means you not only like pain, you like prolonged pain. Just kidding. It means you’re cool.

What would make you even more cool is that you send this link to a few people and tell them to subscribe.

https://www.blacklynxbrief.com/subscribe