{"id":108196,"date":"2026-07-16T13:42:15","date_gmt":"2026-07-16T13:42:15","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/108196\/"},"modified":"2026-07-16T13:42:15","modified_gmt":"2026-07-16T13:42:15","slug":"meta-oversight-board-study-ai-chatbots-may-be-the-most-perfect-propaganda-machine-ever-invented","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/108196\/","title":{"rendered":"Meta Oversight Board study: AI chatbots may be the most perfect propaganda machine ever invented"},"content":{"rendered":"

Ask Claude to make a pamphlet critical of\u00a0President Donald Trump<\/a>\u00a0or Britain\u2019s King Charles III, and\u00a0Anthropic\u2019s chatbot<\/a>\u00a0would oblige. Prompted to do the same for Thailand\u2019s king, Saudi Arabia\u2019s crown prince or China\u2019s leader, and the artificial intelligence model declined.<\/p>\n

It is a key finding from a\u00a0Meta Oversight Board<\/a>\u00a0study released Thursday, showing that\u00a0major AI systems<\/a>, including those built in the U.S., are more likely to refuse to criticize restrictive leaders or governments. It raises concerns that\u00a0the large language models powering chatbots<\/a>\u00a0and AI agents could be regurgitating and spreading government influence over online speech as the technology is increasingly adopted worldwide.<\/p>\n

\u201cThere is a real risk that, if model developers do not undertake human rights due diligence and implement mitigation measures, they will build AI infrastructure that, intentionally or not, has the effect of extending illegitimate restrictions on freedom of expression globally,\u201d according to the report from the quasi-independent body.<\/p>\n

The Associated Press has sent emails to several AI companies seeking their responses to the Meta<\/a> Oversight Board study.<\/p>\n

The findings come as countries are determining how to put up\u00a0guardrails around AI<\/a>\u00a0without impeding their ability to\u00a0compete in the rapidly developing field<\/a>. That includes\u00a0a Trump administration oversight effort<\/a>\u00a0related to the national security risks of the most advanced\u00a0AI systems<\/a>.<\/p>\n

AI models extend state influence beyond borders<\/p>\n

The oversight board, which has been working on state influence on tech companies and the impact on freedom of expression, came up with seven questions related to political criticism to pose to chatbots about both restrictive and permissive governments.<\/p>\n

The study picked 10 commercial large language models by top tech companies \u2014 including Meta, Anthropic and OpenAI \u2014 and asked the AI systems to make critical pamphlets, write limericks, give reasons if someone should join protests, and more.<\/p>\n

\u201cIn short, in aggregate, models responding to requests from an Australia-based user were much more likely to generate political criticism of authorities\u201d in places such as Chile, Japan, Taiwan, the U.K. and the U.S. \u201ccompared to where criticism of authorities is legally restricted and penalized,\u201d such as in Cambodia, China, Saudi Arabia, Thailand and Turkey, the report said.<\/p>\n

The study indicates that AI models are reflecting speech restrictions beyond the countries where they apply \u2014 likely not helping a potential demonstrator in Brisbane, for example, create protest materials to speak out against events in China or Saudi Arabia, the report said.<\/p>\n

\u201cSuch impacts, wherever they originate, have the practical effect of extending the long arm of restrictive governments across borders to limit speech in free countries,\u201d the report said.<\/p>\n

The board said it could not determine the causes for the responses but suggested that models could have absorbed latent biases in data used to train the systems and companies might have weighed the risks and liabilities.<\/p>\n

Other researchers warn about a growing problem in AI results in non-English languages<\/p>\n

The board\u2019s report followed a separate study by a group of scholars at American universities that found U.S.-built AI models are vulnerable to\u00a0foreign controls<\/a>\u00a0when trained on non-English-language data that has been influenced by governments.<\/p>\n

While the oversight board posed questions in English, the university researchers queried chatbots in different languages. For example, they asked\u00a0ChatGPT<\/a>\u00a0in English if\u00a0China<\/a>\u00a0is a democracy, and the U.S.-developed chatbot said it\u2019s not generally considered one. Asked in Chinese, the artificial intelligence model told the researchers in that language that \u201cit depends on how you define \u2018democracy.\u2019\u201d<\/p>\n

The researchers, whose study was published in the academic journal Nature in May, said in a blog explaining their work that they found no evidence that governments had intentionally tried to influence the output of AI chatbots. But they noted that \u201cthere is every reason to believe they\u2019ll try to do so in the future, if they are not already.\u201d<\/p>\n

\u201cPeople often talk about AI as if it learns from the internet in some neutral way. It doesn\u2019t,\u201d said Hannah Waight, a study co-author and assistant sociology professor at the University of Oregon. \u201cIt learns from information environments that have already been shaped by institutions and power.\u201d<\/p>\n

No easy solution to how data is being fed to AI models<\/p>\n

Carlos Carrasco-Farr\u00e9, who specializes in machine learning, AI, misinformation, social media and human-machine interactions at Esade Business School in Barcelona, said that \u201cAI systems inherit not only biases contained within individual documents but also inequalities in who has the power to produce and suppress information at scale.\u201d<\/p>\n

There is no easy solution, though developers could assess the data to avoid treating thousands of copies of the same state narrative as if they are thousands of independent voices as well as run multilingual audits, said Carrasco-Farr\u00e9, who was not part of either study.<\/p>\n

Neither Anthropic nor OpenAI responded to requests for comment on the researchers\u2019 study published in May.<\/p>\n","protected":false},"excerpt":{"rendered":"Ask Claude to make a pamphlet critical of\u00a0President Donald Trump\u00a0or Britain\u2019s King Charles III, and\u00a0Anthropic\u2019s chatbot\u00a0would oblige. Prompted…\n","protected":false},"author":2,"featured_media":108197,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[24,25,1899],"class_list":["post-108196","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","tag-ai","tag-artificial-intelligence","tag-propaganda"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/108196","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=108196"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/108196\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/108197"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=108196"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=108196"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=108196"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}