This week’s edition of The Algorithm highlights a significant shift in the tech landscape, particularly within the realm of artificial intelligence and digital rights. Recent discussions at RightsCon, a conference attended by civil society organizations from around the world, revealed the growing concerns over the loss of funding for global digital rights, especially stemming from the United States government. Eileen Guo, an investigative reporter at MIT Technology Review, underscored the impact of the political climate on international technology policies and operations.
Attendees noted that the Trump administration's rapid cuts to government funding have triggered a reevaluation of the dependency on American tech corporations, especially regarding AI initiatives. This sentiment resonated among European leaders and policymakers, who are increasingly looking to create homegrown alternatives to US-based technologies.
During the conference, Yasmin Curzi, a Brazilian law professor, voiced a strong criticism: “Since Trump’s second administration, we cannot count on [American social media platforms] to do even the bare minimum anymore.” This statement reflects a broader trend where social media platforms are criticized for their failure to effectively moderate content related to gender-based violence. Marlena Wisniak, a human rights lawyer focusing on AI governance, echoed Curzi’s concerns, stating, “The LLMs are moderated poorly, and the poorly moderated LLMs are then also used to moderate other content.”
Many participants argued that the shortcomings in content moderation stem from machine learning models that are primarily trained on English data sets, leading to deficiencies in recognizing and addressing issues in languages and contexts that differ from the Anglo-American baseline. An evaluation of health-care queries processed by ChatGPT indicated that results were significantly less accurate in Chinese and Hindi compared to English and Spanish. This highlights an urgent need for digital platforms to invest in localized solutions.
In response to these concerns, some innovators are developing smaller, community-driven AI models tailored for specific languages and cultures. One such initiative was showcased by Shhor AI at RightsCon, focusing on Indian vernacular languages. These initiatives underscore the potential for smaller language models to outperform larger, multilingual ones in specific, low-resource contexts. Aliya Bhatia, a visiting fellow at the Center for Democracy & Technology, stated that smaller models “might be worthy competitors of multilingual language models in specific, low-resource languages.”
Furthermore, the atmosphere surrounding AI competition was palpable, particularly following the recent Paris AI Summit. The summit emphasized national tech sovereignty, a concept gaining traction in Europe as the continent grapples with its reliance on US tech giants. The European Union, taking bold steps, appointed its first commissioner for tech sovereignty, security, and democracy last November. This appointment reflects a strong desire to create a digital framework internally—one that does not rely on American technology.
In an interesting development, Dutch lawmakers passed motions to disentangle their tech infrastructure from American providers, indicating a wider sentiment across Europe that the current situation is untenable. Andy Yen, CEO of the Switzerland-based digital privacy company Proton, remarked on the trend, saying, “Trump is causing Europe to move faster ... to come to the realization that Europe needs to regain its tech sovereignty.” Such political movements strongly indicate a future where European alternatives to US tech might flourish.
Notably, along with broader implications for AI regulation and governance, OpenAI has also been examining the emotional impacts of its language models. The organization released findings suggesting that female participants who engaged with ChatGPT reported slightly lower socialization rates compared to their male counterparts. This study highlights the complex nuances of AI interactions, particularly how they may shape behaviors and connections among users.
Shifting to another topic, as AI technologies proliferate, risks also increase. A recent case highlighted how malware hidden in AI tools can severely impact individuals. A Disney employee who downloaded such a tool ended up losing his job, illustrating the potential dangers tied to the misuse of AI. The incident serves as a stark reminder of the necessity for ongoing vigilance in tech adoption.
Conversely, privacy concerns remain a critical point of contention. Reports have surfaced that Clearview AI attempted to purchase vast quantities of arrest records and mugshots, including sensitive personal information like social security numbers. This controversial pursuit raises ethical questions regarding data ownership and the implications for privacy rights in the age of AI.
The diverse conversations emerging from RightsCon encapsulate a transformative period in tech governance. As societies strive to establish digital policies grounded in ethical considerations, the interplay between innovation and regulation will play a defining role in shaping the future landscape of AI and digital rights. This moment in tech history is undoubtedly pivotal, ushering in the potential for greater tech sovereignty, local solutions, and a more inclusive approach to AI governance.
