Roundup of the Top AI Stories for Business this Week

Getting your Trinity Audio player ready...

While UK businesses show strong confidence in AI’s potential, actual workplace adoption remains surprisingly low at 23%, with many leaders struggling to implement effective AI strategies. This hesitancy contrasts sharply with the aggressive stance taken by major publishers, as demonstrated by News Corporation’s lawsuit against AI startup Perplexity over unauthorized content scraping – a case that could set important precedents for AI training data rights. Meanwhile, significant technological advances continue, with Anthropic’s Claude AI now capable of performing computer tasks like form filling and website building, though still in experimental stages. In China, AI companies are finding innovative ways to compete globally despite US chip restrictions, with firms like 01.ai achieving impressive results through cost-efficient approaches and specialized training methods. Adding to these developments, Google’s decision to make its AI text watermarking technology, SynthID Text, widely available marks an important step toward addressing concerns about AI content authenticity and transparency, potentially offering a solution to the growing challenge of distinguishing between human and AI-generated content.

1. UK Businesses Want AI But Don’t Know How To Use It
Despite high confidence from UK business leaders in AI’s potential, actual workplace adoption remains low at 23% of workers, while slightly more (25%) use AI in their personal lives. Most business leaders lack a clear implementation strategy, with 83% unsure how to improve efficiency using AI. Worker concerns about job displacement and creativity limitations persist, though younger generations show more openness to AI. The UK ranks among the top countries planning to increase AI investment, with 53% of businesses planning budget increases.

2. News Publishers Sue AI Startup for Stealing Content
News Corporation (representing The Wall Street Journal and New York Post) has sued AI startup Perplexity for copyright infringement, alleging unauthorized content scraping for AI training. The lawsuit seeks $150,000 per infringement and criticizes Perplexity’s “skip the links” approach, which bypasses original sources. News Corp highlighted concerns about AI hallucinations misattributing or fabricating content. While News Corp has paid licensing deals with other AI companies like OpenAI, Perplexity allegedly ignored attempts to negotiate. The case highlights growing tensions between publishers and AI companies over content rights and compensation.

3. New AI Tool Can Do Your Computer Tasks (For Now)
Anthropic has announced that its Claude AI model can now perform computer tasks like moving a mouse cursor, typing text, and completing forms. The experimental feature allows the AI to interact with computer interfaces to handle tasks such as planning trips and building websites. While acknowledging the system is still prone to errors, Anthropic is releasing it early for developer feedback. This development comes as Microsoft launches similar autonomous AI agent capabilities, with both companies positioning these tools as ways to reduce repetitive work rather than replace jobs entirely.

4. Chinese AI Companies Get Creative to Develop Affordable Models
Chinese AI companies are finding innovative ways to compete with Western counterparts despite US chip restrictions. Companies like 01.ai and DeepSeek are reducing costs by using smaller data sets and leveraging China’s affordable engineering talent. 01.ai’s Yi-Lightning model has achieved competitive performance at significantly lower costs than OpenAI’s models. Chinese companies are adopting a “model-of-expert” approach, training multiple specialized neural networks rather than one large model. This strategy, combined with creative data collection methods and cost-efficient operations, allows Chinese AI firms to offer competitive services at lower prices.

5. Google Opens Up AI Text Watermarking Tool to Improve Transparency
Google is making its AI text watermarking technology, SynthID Text, widely available to developers through Hugging Face and its Responsible GenAI Toolkit. The tool embeds invisible watermarks in AI-generated text that persist even after modifications, helping distinguish between human and AI-created content. Originally launched in August 2023 for Google’s Imagen users, this broader release aims to improve transparency in AI content generation and could help advertisers and publishers build trust with their audiences. The timing is significant given growing concerns about AI-generated misinformation and content authenticity.