Meta’s big language model leaked, DuckDuckGo launches its AI-powered search engine, over 60 percent of data breaches are due to remote work.
Welcome to trending hashtags for Tuesday, March 9th.
I’m your host Jim Love, CIO of IT World Canada and TechNewsday in the US – here’s today’s top tech news.
Meta’s latest large language model, LLaMA, has leaked online and is available for download a few weeks after release.
But Meta made its way online anyway via torrent leaks.
Experts warned that similar technology could be used to automate the production of large amounts of fake news, spam, phishing emails, disinformation, provocation, etc.
In fact, Meta said last week that: “More research still needs to be done to address the risks of bias, toxic comments and hallucinations in large language models,” adding that it continues to model the model on a case-by-case basis. Doing- on a sub-case basis to prevent abuse.
However, shortly after it was released, someone on 4Chan published instructions on GitHub on how to obtain the entire model via peer-to-peer file sharing, and eventually download it.
The Register notes that the copies available on GitHub appear to be legitimate. An AI engineer who wrote the download instructions on Microsoft’s code-sharing site showed screenshots of the news site successfully generating text from the model.
While you may be strongly tempted to play with the leaked models, generally, only organizations that have access to GPU stacks and other infrastructure are in a position to build, tweak, and test them. Are. But, according to AI researchers at Meta, LLAMA is built to be smaller and more compact and thus more accessible to academics and developers with smaller computational resources.
Although LLaMA requires hundreds of gigabytes of storage, it mostly takes a fair bit of knowledge to get the model up and running. Reusing it for more nefarious purposes would require further technical expertise.
Meta said it would continue to share the model with select researchers despite the leak.
FBI Director Chris Wray told the Senate Intelligence Committee that the Chinese government has the ability to control software on millions of devices through TikTok.
Senator Marco Rubio asked, “Can they use this to drive a narrative, like dividing Americans against each other?”
Ray replied, “Not only would it be possible, but we’re not sure we’d see many outward signs of it happening, if it were happening”.
He said, “Something that is so sacred in our country – the distinction between the private sector and the public sector – it is a line that does not exist in the Chinese Communist Party.”
The Chinese government owns a 1 percent stake in one of ByteDance’s domestic subsidiaries as well as one of three seats on the subsidiary’s board.
However, TikTok said it operates independently and protects US data through a tie-up with Oracle. Concerns about the company also included previous violations of children’s privacy by the firm.
However, if lawmakers ban the China-based social media platform, Snap, Meta and Google will be the biggest beneficiaries, according to research firm CFRA, adding billions of dollars to TikTok’s advertising revenue.
“Shares of select social media companies could rise on a potential TikTok ban that could gain traction from Congress,” said Angelo Zino, senior equity analyst at CFRA.
Zino explained how the “TikTok ban will move the needle more for SNAP than others” given the similar demographic base using the platforms, adding that growth on YouTube was also accelerated as higher engagement levels And the user base will be attractive to advertisers.
As a matter of fact, Snap shares soared 9.5 percent on Monday after Virginia Senator Mark Warner told Fox News that he would allow the government to ban foreign tech from China.
Although the research firm says that the chances of TikTok getting banned are low now, it has increased significantly in the last six months.
According to Fortinet’s new work-from-anywhere global study, more than 60 percent of companies had a data breach due to remote working vulnerabilities in the past 2 to 3 years.
As part of the study, 570 companies were surveyed. Of which 60 per cent are accommodating employees working from home while 55 per cent are adopting a mixed work strategy.
The high rate of breaches highlights that remote working not only presents theoretical risks, but also risks that are actively being exploited by threat actors to gain access to sensitive information.
The organizations also cited a lack of cybersecurity training for hybrid workers as main concerns, as well as how to extend corporate security to home offices and remote locations.
Organizations are also finding it difficult to implement zero trust access control and deploy patches on devices because of limited visibility on the user’s home environment, branch offices, and off-site locations.
While most organizations have not minimized the risks associated with remote working, it shows that CISOs and security chiefs are investing in new technologies to increase their cyber resilience.
For example, respondents cited network access control, antivirus solutions, multi factor authentication (MFA) and cloud security solutions such as Cloud Access Security Broker (CASB) as the most important tools for securing remote work environments in the future.
The search engine that prides itself on protecting the privacy of searchers and avoiding filter bubbles now wears the tag of artificial intelligence.
DuckDuckGo launched the beta version of an AI search tool called DuckAssist, powered by ChatGPT. The new search engine pulls together information from Wikipedia and related sources such as the Encyclopædia Britannica to generate quick conversational answers to certain queries.
The company said it would limit the data source to Wikipedia to avoid abuse by users of other AI tools.
Unlike other ChatGPT-powered tools we’ve seen flooding in since the beginning of this year, DuckAssist isn’t a chatbot, instead, it suggests an automated answer when it recognizes a search term it can answer. . You are not forced to use the AI search tool, rather you will see a magic wand icon with an “Ask me” button in your search results when an AI-powered response is available.
In addition, the company says it is maintaining privacy by keeping anonymized history in all data shared with users’ searches, and the company’s partners on this project, OpenAI and Anthropic.
The tool is free and available now on the DuckDuckGo web browsing app for phones and computers as well as the company’s browser extension.
That’s the top tech news for today
Links to these stories can be found in articles posted at itworldcanada.com/podcasts. You can also find more great stories and more in-depth coverage at itworldcanada.com or technewsday.com in the US.
Hashtag Trending airs five days a week with the daily newscast and we have a special weekend edition featuring an interview with an expert in some aspect of the technology that is making the news.
We’re always happy to hear from you, you can find me on LinkedIn, Mastodon, Twitter or leave a comment under the article for this podcast at ITWorldCanada.com/podcasts.
I’m your host Jim Love – have a wonderful Thursday!