Tech Beat by Namecheap – 28 July 2023

AI technology has made it possible to mimic people to near perfection, but what could this mean for a society vulnerable to different types of criminal deception? In this week’s lead story, we explore what deepfakes are and learn more about efforts to detect deepfakes in real-time. 

In tech news:

• Generative AI fails at strikebreaking. The ongoing strike in the TV and film industry, coupled with the emergence of AI-based writing and effects, has created a complex environment for show business. However, Fable Studios has chosen an inopportune moment to soft-launch an AI that can fully create a TV show. They have demonstrated this capability with a fake “South Park” episode, according to TechCrunch. While the technology is impressive, with a beginning, middle, and end, and distinct characters, it raises concerns about the future of creative work. The company plans to release a product called Simulation later this year, which uses an agent-based approach to create and document events for media. 

• Billionaire ambitions face a trademark reality check. Elon Musk’s ambitious move to rename Twitter to “X” might land him in a legal quagmire, given the vast number of companies, including giants like Meta and Microsoft, that already possess intellectual property rights for the letter “X.” Nearly 900 active US trademark registrations encompass the letter, according to Reuters. Microsoft has had an “X” trademark since 2003 relating to its Xbox communication, while Meta secured a trademark in 2019 for a blue-and-white “X” linked to software and social media. “There’s a 100% chance that Twitter is going to get sued over this by somebody,” said trademark attorney Josh Gerben.

• Scientists made a ‘biological camera’ that can store images in DNA. Scientists have developed a groundbreaking technique that allows them to capture and store images in DNA, opening up new possibilities for computing and nanotechnology. Vice reports the researchers created a “living digital camera” called BacCam, which uses DNA to encode and retrieve images. By passing light through DNA samples, the researchers were able to create a biological equivalent of a digital camera. This breakthrough offers an alternative method of storing digital data within DNA sequences, without the need for costly and inefficient de novo DNA synthesis. 

• AI titans promise protections and transparency. Seven major AI companies — Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI — have pledged to reduce potential risks associated with artificial intelligence. The BBC reports that as part of their commitment, these companies will undergo security testing by internal and external experts prior to releasing their systems, employ watermarks to make it evident when content is AI-generated, regularly disclose AI capabilities and limitations, and research potential risks such as bias and privacy invasion. These voluntary measures signify progress towards stronger AI regulations in the US and should help the public identify AI-generated online content. 

• Successful robotic transplant surgery. A US surgical team from Washington University School of Medicine in St. Louis successfully conducted the nation’s first robotic liver transplant. As reported in a Wash U press release, this new technique allows surgeons to remotely control a robot’s tools using joystick-like controls. The procedure offers a minimally invasive surgery, with smaller incisions, reduced pain, and quicker recovery. While the first surgery took just over eight hours, similar to conventional liver transplants, further practice is expected to reduce operation time. 

Previously in Tech Beat: Is LinkedIn becoming a hunting ground for spies?

LinkedIn is popular with scammers and spies. Reports of espionage attempts on the platform have surged, with state-backed groups creating fake profiles to target professionals in government, high-tech, and academic fields. The risks of accepting unusual connection requests are real, as spies use LinkedIn for recruitment and phishing scams to steal sensitive data. While the platform employs measures to combat fake profiles, users must remain cautious when engaging with unknown individuals online to protect their information. Read more in our article, LinkedIn’s spy problem.

Tip of the week: Check out ChatBot Arena

In the midst of all this chatbot/generative AI frenzy, LMSYS ORG has introduced Chatbot Arena, a crowdsourced benchmark platform that uses pairwise comparison and the Elo rating system to assess the quality of linguistic models. Users can participate in conversations with anonymous models and vote for their preferred responses. The system has already received over 7,000 votes since its launch a week ago, according to MarkTechPost. So we think it’s time for all of us to give it a try! 

• Navigate to the ChatBot Arena website and select the battle option. 
• Enter text prompts for the chatbots to tackle side-by-side. 
• By providing multiple prompts, you can assess which chatbot performs better. 
• Select the corresponding button to indicate which chatbot is deemed superior or if a tie occurs or both chatbots underperformed. 
• The arena will prompt each chatbot to confirm its identity, leading to surprising results depending on the prompts provided.

While GPT-4 often emerges as a standout, open-source alternatives can also rival it, according to Digital Trends. Although the ChatBot Arena showcases many exciting AI chatbots, it does not include all of them. For those interested in exploring other language models, Bing Chat and Quora’s Poe platform offer intriguing options, such as Bing Chat’s unique personality traits and free access to the Claude+ model, which can compete with GPT-4.

Interested in more of these news updates? Subscribe to Namecheap’s blog today!