Latest Developments in AI with the release of DeepSeek
China’s DeepSeek has been the buzz word all week, with every big tech company responding in their own way. While DeepSeek’s price friendly open-AI service has rocked the industry, it has also raised many concerns about it’s data privacy.
Let’s break this down.
AI has continued to grow exponentially over the last 2 years with billions flooding into AI companies to fund the software. Like ChatGPT, DeepSeek R1 is a "reasoning" model. These models produce responses incrementally, simulating how humans' reason through problems or ideas. OpenAI spent $100 million training GPT-4, while in comparison DeepSeek spent $5 million to develop its model which performs comparably to GPT-4. This figure is still under scrutiny from analysts, however DeepSeek R1’s launch did send a ripple effect globally, with US companies like Nvidia's stock prices dropping significantly; begging the question if the US is over investing AI tech.
Meanwhile, the industry is also engulfed in a debate on whether DeepSeek’s use of data is ethical. DeepSeek’s terms of use state that it collects directly provided data like names and email addresses, automatically collected information like an IP address, and some from other sources such as Apple or Google logins. This information can be stored ‘for as long as possible’ and can be shared with third parties, such as advertisers, analytics companies, law enforcement, public authorities, and copyright holders.
Information requests launched in Italy, Ireland, Belgium, the Netherlands, and France want to know whether the AI company’s collection of information breaches GDPR by transferring personal data to China. The Data Protection Commission (DPC) has written to DeepSeek about Irish users impacted by a data breach discovered last week. It’s the second written request for information the commission has submitted to the Chinese firm in the past week.
What does this mean for Ireland?
While there’s no doubt that DeepSeek’s disruptive launch has proved that innovation is not restricted to just big tech, it also begs the question of ethically handling user data. While there are many open enquiries within Ireland and EU in regard to compliance with GDPR, we would advise not to input any confidential client or personal data into AI models untill there is clear legislation on how a user’s data is processed, stored and shared.
Sources:
DeepSeek: What happened and what does it say about our AI future?