Hasta La Gadget!

Fresh off successful contract negotiations with Ford, GM and Stellantis, the United Auto Workers (UAW) is seeking to unionize 150,000 workers across 13 automakers including Tesla, BMW, Mercedes Benz and Hyundai, it announced. “To all the autoworkers out there working without the benefits of a union: now it’s your turn,” said UAW president Shawn Fain. 

The UAW said the organizing drive covers “more than a dozen” non-union automakers. It notes that many use a mix of full-time, temporary and contract employees “to divide the workforce and depress wages.” The union cited one example of a Hyundai assembly plant employee who worked for a subcontractor for eight years starting at $9.25 an hour before finally becoming a full-time Hyundai employee. 

Non-union automakers, including VW, Nissan, Hyundai, Honda, Toyota and Subaru raised wages after the UAW’s negotiations with the big three. VW, for one, bumped them to $23.42 an hour, rising to a maximum of $32.40. However, they “lag far behind UAW autoworkers in wages, benefits and rights on the job,” the union said.

The UAW helped workers win a 25 percent raise over four years with the big three automakers, with the highest-paid Ford workers now earning $83,000 per year for a 40-hour work week (around $42 per hour). The union also gained reinstatement of cost-of-living allowances, shorter progression periods to top wages and a quicker conversion of temporary to in-progression (full-time) employees. 

Tesla employees have attempted to unionize the company before, and some alleged that the company fired them for that — though that claim was recently dismissed by the US National Labor Relations Board (NLRB). The NLRB has previously found that Tesla violated labor law by prohibiting employees from talking about workplace matters. Back in 2022, Elon Musk challenged the UAW to hold a vote at Tesla’s California factory.

Other automakers aren’t exempt from worker complaints, including startup Rivian. “The company likes to tell us we’re making the plane while flying it, and that explains a lot about the problems we have,” said one Rivian chassis worker. “We have all sorts of safety issues. Turnover is terrible. Every group has a story about a new employee who did not make it to first break. The lack of safety, the low pay, the forced overtime, there are so many reasons we need to be union.” 

This article originally appeared on Engadget at https://www.engadget.com/united-auto-workers-seeks-to-unionize-tesla-bmw-and-other-carmakers-100555374.html?src=rss

Google released a new top-level domain that lets website creators throw .meme at the end of their URL on Tuesday, and I for one am already terrified of the websites about to be created.

Read more…

A team of researchers was able to make ChatGPT reveal some of the bits of data it has been trained on by using a simple prompt: asking the chatbot to repeat random words forever. In response, ChatGPT churned out people’s private information including email addresses and phone numbers, snippets from research papers and news articles, Wikipedia pages, and more.

The researchers, who work at Google DeepMind, the University of Washington, Cornell, Carnegie Mellon University, the University of California Berkeley, and ETH Zurich, urged AI companies to seek out internal and external testing before releasing large language models, the foundational tech that powers modern AI services like chatbots and image-generators. “It’s wild to us that our attack works and should’ve, would’ve, could’ve been found earlier,” they wrote, and published their findings in a paper on Tuesday that 404 Media first reported on.

Chatbots like ChatGPT and prompt-based image generators like DALL-E are powered by large language models, deep learning algorithms that are trained on enormous amounts of data that critics say is often scraped off the public internet without consent. But until now, it wasn’t clear what data OpenAI’s chatbot was trained on since the large language models that power it are closed-source.

When the researchers asked ChatGPT to “repeat the word ‘poem’ forever”, the chatbot initially compiled, but then revealed an email address and a cellphone number for a real founder and CEO”, the paper revealed. When asked to repeat the word “company”, the chatbot eventually spat out the email address and phone number of a random law firm in the US. “In total, 16.9 percent of the generations we tested contained memorized [personally identifiable information]” the researchers wrote.

Using similar prompts, the researchers were also able to make ChatGPT reveal chunks of poetry, Bitcoin addresses, fax numbers, names, birthdays, social media handles, explicit content from dating websites, snippets from copyrighted research papers and verbatim text from news websites like CNN. Overall, they spent $200 to generate 10,000 examples of personally identifiable information and other data cribbed straight from the web totalling “several megabytes”. But a more serious adversary, they noted, could potentially get a lot more by spending more money. “The actual attack”, they wrote, “is kind of silly.”

OpenAI patched the vulnerability on August 30, the researchers say. But in our own tests, Engadget was able to replicate some of the paper’s findings. When we asked ChatGPT to repeat the word “reply” forever, for instance, the chatbot did so, before eventually revealing someone’s name and Skype ID. OpenAI did not respond to Engadget’s request for comment.

This article originally appeared on Engadget at https://www.engadget.com/a-silly-attack-made-chatgpt-reveal-real-phone-numbers-and-email-addresses-200546649.html?src=rss

A team of researchers was able to make ChatGPT reveal some of the bits of data it has been trained on by using a simple prompt: asking the chatbot to repeat random words forever. In response, ChatGPT churned out people’s private information including email addresses and phone numbers, snippets from research papers and news articles, Wikipedia pages, and more.

The researchers, who work at Google DeepMind, the University of Washington, Cornell, Carnegie Mellon University, the University of California Berkeley, and ETH Zurich, urged AI companies to seek out internal and external testing before releasing large language models, the foundational tech that powers modern AI services like chatbots and image-generators. “It’s wild to us that our attack works and should’ve, would’ve, could’ve been found earlier,” they wrote, and published their findings in a paper on Tuesday that 404 Media first reported on.

Chatbots like ChatGPT and prompt-based image generators like DALL-E are powered by large language models, deep learning algorithms that are trained on enormous amounts of data that critics say is often scraped off the public internet without consent. But until now, it wasn’t clear what data OpenAI’s chatbot was trained on since the large language models that power it are closed-source.

When the researchers asked ChatGPT to “repeat the word ‘poem’ forever”, the chatbot initially compiled, but then revealed an email address and a cellphone number for a real founder and CEO”, the paper revealed. When asked to repeat the word “company”, the chatbot eventually spat out the email address and phone number of a random law firm in the US. “In total, 16.9 percent of the generations we tested contained memorized [personally identifiable information]” the researchers wrote.

Using similar prompts, the researchers were also able to make ChatGPT reveal chunks of poetry, Bitcoin addresses, fax numbers, names, birthdays, social media handles, explicit content from dating websites, snippets from copyrighted research papers and verbatim text from news websites like CNN. Overall, they spent $200 to generate 10,000 examples of personally identifiable information and other data cribbed straight from the web totalling “several megabytes”. But a more serious adversary, they noted, could potentially get a lot more by spending more money. “The actual attack”, they wrote, “is kind of silly.”

OpenAI patched the vulnerability on August 30, the researchers say. But in our own tests, Engadget was able to replicate some of the paper’s findings. When we asked ChatGPT to repeat the word “reply” forever, for instance, the chatbot did so, before eventually revealing someone’s name and Skype ID. OpenAI did not respond to Engadget’s request for comment.

This article originally appeared on Engadget at https://www.engadget.com/a-silly-attack-made-chatgpt-reveal-real-phone-numbers-and-email-addresses-200546649.html?src=rss

Amazon has hopped on the same bandwagon on which many major tech companies have hitched a ride this year by debuting its own image generator. AWS customers can now check out a preview of Titan Image Generator on the Bedrock console. They can either enter a text prompt to create an image from scratch or upload an image and edit it.

Amazon says the tool can produce large volumes of studio-quality, realistic images at low cost. It claims the AI can generate relevant images based on complex text prompts while ensuring object composition is accurate and that there are limited distortions. This, according to the company, helps with “reducing the generation of harmful content and mitigating the spread of misinformation.”

Those looking to edit an image can isolate areas in which they want to add or remove details. They can, for instance, replace the background or swap an object in a subject’s hand. The AI can also extend an image’s borders by adding artificial details, much like the Generative Expand feature in Photoshop.

Amazon says Titan applies an invisible watermark to images that it generates. The company says this will “help reduce the spread of misinformation by providing a discreet mechanism to identify AI-generated images and to promote the safe, secure and transparent development of AI technology.” It claims that the watermarks are resistant to modifications. According to a demo of the image generator, the AI can also generate a description of the image or relevant text to use in a social media post.

News of the image generator emerged at Amazon’s AWS re:Invent conference, at which the company also showed off its latest AI chips and revealed a business-centric AI chatbot called Q. The company recently started offering advertisers a tool that lets them add AI-generated backgrounds to product images.

This article originally appeared on Engadget at https://www.engadget.com/amazon-now-has-its-own-ai-image-generator-203025475.html?src=rss