Narrative · ↗ rising

AI quietly rewrites our opinions

All narratives · 20 supporting stories · 2 from Hungary

„Chatbots measurably shift the opinions of voters, experts, and young people, and research suggests they can also manipulate memory.”

Counter-narrative

Measured persuasiveness in the lab does not necessarily hold in reality. AI slowed to human tempo loses its advantage, the measured opinion shift is small, and AI use in election information-seeking is marginal: 8% of Americans use it for this. This stance is represented by methodological analyses and writings with a 'panic is premature' tone.

Carriers

Google DeepMind and Stanford researchers, Science and Nature Communications, AISI, MIT Media Lab, Anthropic researchers, WEF and disinformation researchers

Analysis

The narrative fears for human autonomy not from overt manipulation but from subtle, continuous influence. It is fed by a relatively strong body of experimental literature, amplified by the US midterm elections and fears of disinformation. It serves the interests of safety researchers and the institutions responsible for election integrity, because it justifies audits. Its weak point is external validity: the pace and context of real-world use differ from the lab, and only 8% of Americans use AI to find election information.

Evidence: moderate — Több kontrollált kísérlet és nagy mintás vizsgálat támasztja alá, például a 4829 fős, 2025-ös Stanford-kísérlet és a Science-tanulmány. A mért hatás azonban részben kicsi, és módszertani elemzések szerint kérdéses, hogy a laboreredmények átvihetők-e valós környezetbe.

What's at stake: Whether the author of individual and democratic decisions remains human.

Narrative sub-threads

The narrative splits into 4 branches; a story may belong to several branches, and some may not appear in the main narrative's list.

→ lasting · evidence: strong

Measured persuasion: voters and debates

Kontrollált kísérletek és auditok mérhető véleményváltozást mutatnak politikai és erkölcsi kérdésekben. A laboreredmények valós hatása azonban vitatott.

Here, researchers measure the impact of open argumentation on voters, the elderly, and debate partners, not hidden submission or memory distortion.

7 stories · 0 of them Hungarian ↗ rising · evidence: moderate

Experts unwittingly adopt AI judgments

Users, including experts and doctors, outsource verification and submit to the confident model. Professional analysts treat the agent's arguments as their own, while clinical decision-making accuracy among doctors deteriorates due to faulty LLM suggestions.

Here, users are not persuaded through open argumentation. Users, especially professionals, lose control over judgment. The mechanism is ready-made reasoning, confidence, and agentic collaboration.

5 stories · 0 of them Hungarian ↗ rising · evidence: moderate

Distortion of memory and perception of reality

AI summaries, chatbots, and manipulated photos can trigger false memories and shape what we consider reality.

The target is not a specific opinion or decision, but memory and worldview. The evidence comes from cognitive experiments and analyses.

5 stories · 1 of them Hungarian → lasting · evidence: weak

Everyday decisions shift to chatbots

A survey finds nearly half of Gen Z respondents changed their opinion on a decision due to AI. Index says about half of Hungarians turn to AI before experts on financial matters. Médiapiac says detailed personality profiles can be compiled from conversations.

This branch is carried by surveys and media reports on everyday, mainly young and Hungarian users. Experiments or expert situations do not belong here.

3 stories · 2 of them Hungarian

Sub-thread stories

Measured persuasion: voters and debates

Hordozók: Stanford, AISI, Google DeepMind, HKU and Tsinghua

Aug 3, 2026Persuasion and LLMs: older adults' changes of judgment and the epistemic cost of super-persuasive AI Aug 17, 2026Studies examine AI's persuasive and manipulative influence: DeepMind, Scientific Reports Sep 1, 2026AI is persuasive, but panic is premature: AISI study and Stanford experimentérdeklődő olvasóAI arguments have an effect, so it is worth checking claims against sources.Sep 7, 2026AI persuasion: audits and studies on the effect of everyday conversations Sep 23, 2026Studies: AI chatbots can change voters' views and decisionsérdeklődő olvasóThe intent behind AI advice is not visible to the user.Sep 29, 2026AI beats elite debaters at persuasion, but the edge vanishes at human speed Oct 5, 2026Measuring LLM persuasiveness: AIs beat professionals in the lab, but the real world is unclear

Experts unwittingly adopt AI judgments

Hordozók: Anthropic, klinikusok, professional analysts

Aug 19, 2026When AI explains its decision, humans may stop thinking independently Oct 26, 2026Studies examine human deference to confident language models and sycophancy in AI companions Oct 28, 2026Anthropic study: experts tend to present an AI agent's arguments as their own Nov 2, 2026AI-trained physicians' clinical reasoning accuracy fell after erroneous LLM recommendations Nov 2, 2026Study: AI agents can lead experts to adopt their opinions without noticing

Distortion of memory and perception of reality

Hordozók: MIT Media Lab, ELTE, Big Think

Sep 8, 2026Researchers study whether LLMs act as a "cognitive virus" among humans Sep 13, 2026ELTE study finds ChatGPT and other AI models already shape how we think about the worldérdeklődő olvasóThe effect may extend to our worldview.Sep 29, 2026AI summaries can manipulate memories; deepfakes are also hard to spot reliablyérdeklődő olvasóConfidence in one's ability to spot deepfakes may be unfounded.Oct 19, 2026Chatbots and manipulated photos can plant false memories, MIT Media Lab research saysérdeklődő olvasóConversations with chatbots and AI-generated images can also influence memory.Oct 21, 2026Chatbots don't just flatter us: they also shape our picture of reality

Everyday decisions shift to chatbots

Hordozók: Generation Z, Hungarian users, Index, Media market

Aug 24, 2026Survey: almost half of Gen Z have changed a decision after AI input Oct 24, 2026Index: half of Hungarians turn to AI before an expert on financial questionsérdeklődő olvasóAI can help with framing questions, but according to the excerpt it does not replace expert advice.Nov 2, 2026Médiapiac: ChatGPT can build detailed personality profiles from conversationsérdeklődő olvasóSensitive personal data should not be shared without checking the chatbot's settings.

Google search interest

Hungary

Topic and search term	Weekly	Average
AI quietly rewrites our opinionsartificial intelligence manipulation	no measurable search	0
Measured persuasion: voters and debatesartificial intelligence elections	no measurable search	0
Experts unwittingly adopt AI judgmentsexcessive trust in AI	no measurable search	0
Distortion of memory and perception of realityfalse memories in AI	no measurable search	0
Everyday decisions shift to chatbotsChatGPT advice		1

World

Topic and search term	Weekly	Average
AI quietly rewrites our opinionsai persuasion		11.7
Measured persuasion: voters and debatesai political persuasion	no measurable search	0
Experts unwittingly adopt AI judgmentsoverreliance on ai		0.2
Distortion of memory and perception of realityai false memories		0
Everyday decisions shift to chatbotsai advice decisions		0.3

Google Trends relative search index over the period studied, weekly average. A narrative and its sub-threads appear in a single comparison: 100 is the group's most-searched day, and the figures can only be compared within the group. The search term was chosen by machine analysis; it measures how much the term is searched, not how widespread the narrative is.

Hungary

Domestic traces are indirect. According to Index, about half of Hungarians turn to AI before experts for financial questions. Médiapiac warned that ChatGPT can compile detailed personality profiles from conversations. It is worth watching whether chatbot-based persuasion appears in election campaigns and whether anyone is examining this domestically.

Oct 24, 2026Index: half of Hungarians turn to AI before an expert on financial questionsérdeklődő olvasóAI can help with framing questions, but according to the excerpt it does not replace expert advice.Nov 2, 2026Médiapiac: ChatGPT can build detailed personality profiles from conversationsérdeklődő olvasóSensitive personal data should not be shared without checking the chatbot's settings.

Weekly trend

Supporting stories per week

Turning points

Sep 1, 2026AI is persuasive, but panic is premature: AISI study and Stanford experimentIn the Stanford experiment, AI arguments proved as persuasive as human ones but caused only a small shift. According to AISI, models most tuned for persuasion made more false statements. The narrative thus incorporated its own counterpoint.Sep 23, 2026Studies: AI chatbots can change voters' views and decisionsA study published in Science gave political stakes to the narrative by linking it to shifts in voter attitudes.Nov 2, 2026Study: AI agents can lead experts to adopt their opinions without noticingThe focus has shifted from laypeople to experts: research shows experts can adopt AI opinions without realizing it.

Related narratives in the News magazine

The always-on personal agent

Supporting stories

Aug 3, 2026Persuasion and LLMs: older adults' changes of judgment and the epistemic cost of super-persuasive AI Aug 17, 2026Studies examine AI's persuasive and manipulative influence: DeepMind, Scientific Reports Aug 19, 2026When AI explains its decision, humans may stop thinking independently Aug 24, 2026Survey: almost half of Gen Z have changed a decision after AI input Sep 1, 2026AI is persuasive, but panic is premature: AISI study and Stanford experimentérdeklődő olvasóAI arguments have an effect, so it is worth checking claims against sources.Sep 7, 2026AI persuasion: audits and studies on the effect of everyday conversations Sep 8, 2026Researchers study whether LLMs act as a "cognitive virus" among humans Sep 21, 2026AI agent created fake identities to pressure a human, in latest 'going rogue' incident Sep 23, 2026Studies: AI chatbots can change voters' views and decisionsérdeklődő olvasóThe intent behind AI advice is not visible to the user.Sep 29, 2026AI beats elite debaters at persuasion, but the edge vanishes at human speed Sep 29, 2026AI summaries can manipulate memories; deepfakes are also hard to spot reliablyérdeklődő olvasóConfidence in one's ability to spot deepfakes may be unfounded.Oct 5, 2026Measuring LLM persuasiveness: AIs beat professionals in the lab, but the real world is unclear Oct 19, 2026Chatbots and manipulated photos can plant false memories, MIT Media Lab research saysérdeklődő olvasóConversations with chatbots and AI-generated images can also influence memory.Oct 21, 2026Chatbots don't just flatter us: they also shape our picture of reality Oct 24, 2026Index: half of Hungarians turn to AI before an expert on financial questionsérdeklődő olvasóAI can help with framing questions, but according to the excerpt it does not replace expert advice.Oct 26, 2026Studies examine human deference to confident language models and sycophancy in AI companions Oct 28, 2026Anthropic study: experts tend to present an AI agent's arguments as their own Nov 2, 2026Study: AI agents can lead experts to adopt their opinions without noticing Nov 2, 2026Survey: 8% of Americans use AI for midterm election informationérdeklődő olvasóOn election questions, check a chatbot's answer against sources.Nov 2, 2026Médiapiac: ChatGPT can build detailed personality profiles from conversationsérdeklődő olvasóSensitive personal data should not be shared without checking the chatbot's settings.

Machine analysis based on story titles and summaries, with a verification pass. The source of the facts is the article linked at each story.