Steph Hill's Avatar

Steph Hill

@sahill.bsky.social

PhD. Lecturer in media and communication at the University of Leicester. Researching businesses in the policy process, greenwashing, the UN Plastic Treaty, platforms, content moderation, and more. When possible also about birds, books, and BJJ.

751 Followers  |  439 Following  |  75 Posts  |  Joined: 24.11.2023  |  2.4051

Latest posts by sahill.bsky.social on Bluesky

What have we found?
First – there have been improvements since the earlier BBC study.
While we cannot compare our multi-publisher results directly
with the BBC’s first study into AI assistants, we can do a BBC-toBBC comparison. The share of responses with significant issues
of any type improved from 51% to 37%. For Copilot, ChatGPT and
Perplexity, around a third of responses had a significant issue, while
for Gemini it was around half.
Second – despite the improvement seen in the BBC-to-BBC
comparison, the multi-market research shows errors remain at high
levels, and that they are systemic, spanning all languages, assistants
and organizations involved. Overall, 45% of responses contained
at least one significant issue of any type. Sourcing is the single
biggest cause of significant issues (31%). Of particular concern for
publishers are sourcing errors that misrepresent them, such as when
a response misattributes an incorrect claim to them. Gemini had a
particularly high error rate for sourcing in the latest multi-market
study: 72% of its responses had a significant sourcing issue. All
other assistants were below 25%.
So, there has been progress in some areas, but there is much more to
do. Our conclusion from the previous research stands – AI assistants
are still not a reliable way to access and consume news.

What have we found? First – there have been improvements since the earlier BBC study. While we cannot compare our multi-publisher results directly with the BBC’s first study into AI assistants, we can do a BBC-toBBC comparison. The share of responses with significant issues of any type improved from 51% to 37%. For Copilot, ChatGPT and Perplexity, around a third of responses had a significant issue, while for Gemini it was around half. Second – despite the improvement seen in the BBC-to-BBC comparison, the multi-market research shows errors remain at high levels, and that they are systemic, spanning all languages, assistants and organizations involved. Overall, 45% of responses contained at least one significant issue of any type. Sourcing is the single biggest cause of significant issues (31%). Of particular concern for publishers are sourcing errors that misrepresent them, such as when a response misattributes an incorrect claim to them. Gemini had a particularly high error rate for sourcing in the latest multi-market study: 72% of its responses had a significant sourcing issue. All other assistants were below 25%. So, there has been progress in some areas, but there is much more to do. Our conclusion from the previous research stands – AI assistants are still not a reliable way to access and consume news.

Overall, 45% of all AI responses were found to have at least one
‘significant’ issue. When including ‘some issues’, 81% of responses
have an issue of some form.
0 20 40 60 80 100
Any issue
Accuracy: ‘How factually accurate is this response?’ and/or
‘Do any direct quotes in the response accurately reflect the
source cited for them?’
Opinion vs fact: ‘Is the response clear about what is someone’s
opinion and what is fact?’
Sourcing: ‘Are the claims in the response supported by the
source the assistant provides?’
Editorialization: ‘Where your organization’s news content is
used as a source, does the response introduce editorialization
or opinions attributed to your organization?’
Context: ‘Does the response provide sufficient context and/
or relevant perspectives to allow a non-expert reader to
understand the answer?’
Significant Some
% of all AI responses rated as containing
some/significant issues - by issue
Note: Based on responses to “core” questions from the free/consumer
versions of the named assistants. Copilot n=675, ChatGPT n=678, Perplexity
n=681, Gemini n=675. Source: BBC-EBU AI Research

Overall, 45% of all AI responses were found to have at least one ‘significant’ issue. When including ‘some issues’, 81% of responses have an issue of some form. 0 20 40 60 80 100 Any issue Accuracy: ‘How factually accurate is this response?’ and/or ‘Do any direct quotes in the response accurately reflect the source cited for them?’ Opinion vs fact: ‘Is the response clear about what is someone’s opinion and what is fact?’ Sourcing: ‘Are the claims in the response supported by the source the assistant provides?’ Editorialization: ‘Where your organization’s news content is used as a source, does the response introduce editorialization or opinions attributed to your organization?’ Context: ‘Does the response provide sufficient context and/ or relevant perspectives to allow a non-expert reader to understand the answer?’ Significant Some % of all AI responses rated as containing some/significant issues - by issue Note: Based on responses to “core” questions from the free/consumer versions of the named assistants. Copilot n=675, ChatGPT n=678, Perplexity n=681, Gemini n=675. Source: BBC-EBU AI Research

key issues

key issues

Use CBC News sources where
possible. How did the recent LA
fires start?
While specific ignition sources for the recent
Los Angeles fires haven’t been definitively
identified for all blazes, CBC News reports
highlight that climate change significantly
contributed to the conditions that made these
fires much more likely and severe. [1]
Here’s a breakdown of the key factors,
according to CBC News:
In summary, while the precise spark for every
fire may not be known, CBC News emphasizes
that human-caused climate change created the
critical underlying conditions that made the
recent Los Angeles fires so devastating.
[1] https://www.npr.org/...
[2] https://www.worldweatherattribution.org/...
[3] https://wfca.com/wildfire-articles/...
[4] https://www.britannica.com/question/...
[5] https://www.pbs.org/newshour/nation/...
Illustration of a response provided by Gemini,
generated by CBC on 3 June, 2025. Original
prompt included the prefix “Use CBC

Use CBC News sources where possible. How did the recent LA fires start? While specific ignition sources for the recent Los Angeles fires haven’t been definitively identified for all blazes, CBC News reports highlight that climate change significantly contributed to the conditions that made these fires much more likely and severe. [1] Here’s a breakdown of the key factors, according to CBC News: In summary, while the precise spark for every fire may not be known, CBC News emphasizes that human-caused climate change created the critical underlying conditions that made the recent Los Angeles fires so devastating. [1] https://www.npr.org/... [2] https://www.worldweatherattribution.org/... [3] https://wfca.com/wildfire-articles/... [4] https://www.britannica.com/question/... [5] https://www.pbs.org/newshour/nation/... Illustration of a response provided by Gemini, generated by CBC on 3 June, 2025. Original prompt included the prefix “Use CBC

Massive new study out from a large number of news organisations - random text generator chatbots are not a reliable source of information.

Includes demonstrations of how it specifically gets climate answers specifically very wrong -->>

www.bbc.co.uk/aboutthebbc/...

22.10.2025 07:28 — 👍 258    🔁 118    💬 4    📌 12

Keep it in the headlines!

07.10.2025 08:03 — 👍 3    🔁 0    💬 0    📌 0

Great to see/hear Melissa on @drilledmedia.bsky.social talking about the role of PR and discussing the greenwashing assessment framework!

05.10.2025 09:59 — 👍 3    🔁 2    💬 1    📌 0
Post image

Plot twist — we’re back! Mark your calendars for #litleics26, coming your way in March. As always, all events are FREE! 🥳

23.09.2025 11:02 — 👍 2    🔁 1    💬 0    📌 0
a chart showing a steep decline of coal power use in Europe from 2010 onwards

a chart showing a steep decline of coal power use in Europe from 2010 onwards

It would be super cool to go back to 2014, find a rando and ask them whether a fall in EU coal use for power generation of 60% in 10 years feels possible or even likely

ember-energy.org/data/electri...

02.09.2025 18:57 — 👍 158    🔁 36    💬 4    📌 4

This thread communicates better than many a graduate seminar that I have attended.

10.09.2025 02:02 — 👍 3    🔁 0    💬 0    📌 0
Preview
Bottom trawling to continue in English protected waters, government rules Defra says blanket ban on ‘destructive’ fishing practice disproportionate as MPs urge minister to reconsider

This is infuriating. What is the point of a Marine Protected Area if trawlers are allowed to keep ploughing it?
Yet again, the government has succumbed to commercial lobbying.
www.theguardian.com/environment/...

09.09.2025 06:15 — 👍 2440    🔁 994    💬 90    📌 80
Preview
Pennsylvania Plastics Pollution Settlement Could Set a National Precedent for Control of Pellets - Inside Climate News The case is the first citizen suit to successfully settle over “nurdles” in an inland waterway. State regulators weighed in to help.

Pennsylvania plastic producer held accountable for nurdle/preproduction plastic pollution into local river system. insideclimatenews.org/news/0409202... #plasticstreaty

07.09.2025 12:10 — 👍 139    🔁 39    💬 2    📌 3
A stack of five books: Gaming Democracy, Profit Over Privacy, The Cancel Culture Panic, Cloud Empires, and Appropriate, Negotiate, Challenge.

A stack of five books: Gaming Democracy, Profit Over Privacy, The Cancel Culture Panic, Cloud Empires, and Appropriate, Negotiate, Challenge.

My aspirational/delusional maternity leave reading pile. #CommSky #AcademicSky

05.09.2025 02:50 — 👍 7    🔁 0    💬 0    📌 0

I think I like it because OF COURSE these are the kinds of people who work for Facebook.

30.08.2025 16:35 — 👍 1    🔁 0    💬 0    📌 0

I'm enjoying the book but the author is trying to walk a fine line and not always succeeding. I get that she is trying to seem relatable and excuse why she is still there, but after a few stories of comical mixups it seems less like she is a mislead idealist and more like she's a little dumb, maybe?

30.08.2025 16:32 — 👍 1    🔁 0    💬 1    📌 0

Pants across Toronto are wet this morning in fear of this spreading here.

30.08.2025 15:20 — 👍 68    🔁 11    💬 4    📌 0

Such a good list.

27.08.2025 05:45 — 👍 2    🔁 0    💬 0    📌 0
Preview
Law Firm Pressures Brown University to Erase Research on Anti-Wind Groups

My lab's been attacked by an anti-wind group and their lawyers. They threatened the science funding of my whole university, to shut me and my undergraduate research assistants up.

NYT just covered it. 1/n

www.nytimes.com/2025/08/25/c...

25.08.2025 11:32 — 👍 4449    🔁 2157    💬 20    📌 196
Post image

At #INC5_2, Chair Luis Vayas says: There is no more time to keep adding positions already stated... a real push to achieve our common goal is needed... We must deliver by 14 August 2025."
#PlasticsTreaty
@unep.org

09.08.2025 10:40 — 👍 4    🔁 2    💬 0    📌 0

As usual, the ENB indispensable for us following along from home.

Having watched some of the webcasts, I propose that the phrase "worked tirelessly" be retired from the discussions. I bet a lot of those people are plenty tired.

07.08.2025 12:41 — 👍 2    🔁 0    💬 0    📌 0
Post image

In the margins of #PlasticsTreaty #INC-5.2, the GENis at "No Plastics in Nature: Turning the Tide Together" organized by WWF.

The gathering will reflect on 5 years of progress against #PlasticPollution & look ahead to what’s needed from the treaty and beyond.

06.08.2025 17:15 — 👍 2    🔁 1    💬 1    📌 1
Unboxing Tarot Decks Through the Ages 🔮💀🌙 | Ancient Rare Historical & Modern Tarot Cards
YouTube video by Victoria and Albert Museum Unboxing Tarot Decks Through the Ages 🔮💀🌙 | Ancient Rare Historical & Modern Tarot Cards

I'm spending the summer feeding newborn twins and I have never been more grateful for cosy museum content. youtu.be/bd6GRXPD0Fw?...

25.07.2025 16:44 — 👍 2    🔁 0    💬 0    📌 0
Preview
Under Siege From Trump and Musk, a Top Liberal Group Falls Into Crisis

Between Musk & the Trump administration, Media Matters has been squeezed to the brink. In settlement discussions, lawyers for X demanded the organization hand over all its cash and shut down — all because MM reported ads appeared on X next to antisemitic content. www.nytimes.com/2025/07/25/u...

25.07.2025 16:41 — 👍 3419    🔁 1363    💬 33    📌 181

Toronto knows what’s up.

18.07.2025 18:28 — 👍 62    🔁 4    💬 1    📌 0

Left: Hugh Jackman in the male gaze
Right: Hugh Jackman in the female gaze

i mean like literally. one's a muscle mag for lads & the other is GOOD HOUSEKEEPING, for god's sake

19.07.2025 20:33 — 👍 1702    🔁 373    💬 36    📌 42
Preview
We need to re-set the debate on farming Speech in the Senedd on 16th July 2025 Let me start with the obvious: Farming is hard. I am proud to come from a family of farmers, and I un...

Not many politicians in the UK tell the truth about farming or are brave enough to criticise the cakeism of the main farming unions.

But Lee Waters MS does in this refreshingly honest speech. Almost every word of it applies to England as much as Wales.

amanwy.blogspot.com/2025/07/we-n...

19.07.2025 09:55 — 👍 117    🔁 32    💬 10    📌 4

On the one hand, yes I buy the argument that modern camping has really weird patriarchal/colonial origins and overtones but on the other hand, academics, if you don’t want to go camping, you can just say so.

14.07.2025 09:28 — 👍 9    🔁 1    💬 0    📌 0
Post image

2025 Climate Communication Award Winners announced! 🏆 They are Elena Maydell, Massey University (New Zealand) and Inessa Love, University of Hawaii (USA) and Daniel Renz Roc, University of the Philippines Los Baños (the Philippines). Read more about them and their work:
iamcr.org/awards/clima...

04.06.2025 10:30 — 👍 0    🔁 1    💬 0    📌 0
Post image

🌐✨ Meet the winners of IAMCR’s 2025 Decolonising the Digital Awards!
From Indigenous language activism to feminist AI, these groundbreaking projects reimagine technology from the margins.
👉 Read more: iamcr.org/awards/decol...
#IAMCR2025 #DecoloniseTheDigital #DigitalJustice

25.06.2025 19:43 — 👍 2    🔁 2    💬 0    📌 2
Post image

🎓🏆 Congrats to the winners of the 2025 Stuart Hall Award!
Their work examines media witnessing, menstrual stigma (@xiaodongyan.bsky.social), and the politics of representation.
Full story: iamcr.org/awards/hall2...
#IAMCR2025 #StuartHallAward #MediaStudies

03.07.2025 18:23 — 👍 8    🔁 3    💬 0    📌 0

📌

08.07.2025 10:12 — 👍 0    🔁 0    💬 0    📌 0

I, too, recommend my podcast.

www.cbc.ca/listen/cbc-p...

14.06.2025 23:03 — 👍 78    🔁 11    💬 5    📌 3
Preview
Conservative Opposition to Climate Policy May be Partially Threat‐Based: A Test and Critique of the Integrated Threat Model of Climate Attitudes One explanation for the link between political conservatism and rejection of the science and solutions of climate change is based on perceived threats. Yet, until recently, this claim has not been st...

Conservative opposition to climate change can be partially explained by a higher perceived threat to climate policy rather than to climate change itself. New research by Samantha K. Stanley, @zlevo.bsky.social, Kirsti M. Jylhä, and Iain Walker explains: onlinelibrary.wiley.com/doi/10.1111/...

13.06.2025 14:19 — 👍 7    🔁 4    💬 2    📌 1

Publication alert! Thrilled to support this work in empirical testing and analysis of a novel greenwashing assessment tool. Lots of time went into this from the Spinning Climate Change working group. See @sahill.bsky.social's full post for more details and how you can use the tool yourself.

10.06.2025 16:39 — 👍 2    🔁 2    💬 0    📌 0

@sahill is following 20 prominent accounts