CheckMate is a weekly newsletter from RMIT FactLab recapping the latest in the world of fact checking and misinformation. It draws on the work of FactLab's researchers and journalists, including its CrossCheck unit, and of its sister organisation, RMIT ABC Fact Check.
You can subscribe to have the next edition delivered straight to your inbox.
CheckMate May 5, 2023
This week, CheckMate tackles a claim by Senator for Queensland Matt Canavan that shutting down the Liddell power station led to an electricity shortfall.
We also bring you tips on how to spot AI-generated images, and reveal the state of fact checking around the world.
Did Liddell's closure lead to a 'lack of power' in NSW?
Just hours after the Liddell power station's final unit was powered down late last week, Nationals senator Matt Canavan took to social media to suggest the closure had already caused issues for Australia's National Electricity Market (NEM).
"First day without the Liddell coal power station and the notices of a lack of power are flying thick and fast," the senator wrote on Twitter. "Not a good sign and it is not even winter yet."
Included in his tweet were four screenshots of announcements published to the Australian Energy Market Operator's (AEMO) website, three of which forecast a "lack of reserve" (LOR) for electricity capacity in NSW, while the fourth warned of a possible market intervention in South Australia.
According to AEMO, a LOR1 forecast, like those shared by Senator Canavan, "signals a reduction in predetermined electricity reserve levels" and "simply provides an indication to the market to encourage more generation".
"At this level, there is no impact to power system security or reliability," it states.
The notice flagging possible market intervention for South Australia, meanwhile, was issued due to "synchronous generating units" that were predicted to be "inadequate to maintain a secure operating state".
The notice stated that "in the absence of sufficient market response", AEMO "may need to intervene by issuing a direction" that would force some electricity generators to operate (or remain "synchronised").
But while all four notices shared by Senator Canavan were issued on the day of Liddell's closure (April 28), was he right to link AEMO's warnings to the event?
Alison Reeve, the deputy director of the Grattan Institute's energy and climate program, told CheckMate it was "very unlikely the LOR1 notices were linked to the closure of Liddell".
Similarly, Bruce Mountain, director of the Victorian Energy Policy Centre at Victoria University, told CheckMate that neither the LOR1 forecasts nor the market intervention notice for South Australia could be linked to the closure of Liddell.
Both experts said that such notices were common.
Indeed, as reports published by AEMO show, there were 19 LOR1 forecasts issued during the quarter ending in March 2023, 45 in the December 2022 quarter and 89 in the September 2022 quarter.
Moreover, in reports for previous June quarters — the equivalent time periods for the notices shared by Senator Canavan — there were 144 LOR1 forecasts in 2022 (or 23 excluding the period when there was an extraordinary market suspension) and 24 in 2021.
According to Ms Reeve, LOR1 notices were "issued when the amount of available generation capacity above forecast demand (the reserve) is less than the capacity of the two largest units in a state".
As she put it, such notices were the equivalent of AEMO saying: "If the two biggest power stations in this state shut down in the next 30 minutes, the lights would go out. Can someone please send more electricity out."
For its part, AEMO general manager of system design Merryn York told the ABC in the days prior to Liddell's closure: "The notification of Liddell's retirement has allowed the market to respond, with NSW forecast to meet reliability measures until at least 2025."
Professor Mountain, meanwhile, noted that in its final weeks Liddell had been producing "much more [electricity] than it had often in the last few years, presumably to burn through spare coal reserves".
In an email, he explained that the plant's closure did mean there had been a "reasonable reduction in dispatchable capacity" but added that there was "more than enough left for this to be a non-issue".
"The bigger problem is the overall poor reliability of coal generation," Professor Mountain said, quipping that "Senator Canavan would be correct in as much as he is wishing to draw attention to the importance of quickly replacing Liddell (and the remaining coal generators) with renewable generation and storage."
Notably, of the four AEMO notices posted to Twitter by Senator Canavan, three were subsequently cancelled by the operator.
Struggling to spot the fakes? Here are some tips
With the sudden rise of photorealistic images generated by artificial intelligence (AI), separating genuine photos from fakes is only getting harder.
There is, sadly, no foolproof way to weed out artificial images, but there are still plenty of ways to spot them.
In a helpful article by fact checkers with AFP, experts explained that AI draws on potentially millions of images, "deconstructs them and then reconstructs a photo pixel by pixel, which means that in the final rendering, we no longer notice the difference between the original images".
Compounding the difficulties of verification, they added, images posted to social media are typically stripped of metadata, or information that may otherwise point to their origin.
So, what can be done?
According to the fact checkers, a good start is finding the first instance of an image, since its creator may have already flagged it as being AI-generated.
These can be found using reverse image searching tools such as Google, Yandex or TinEye, which may also reveal similar pictures from reliable sources that the image can be compared against.
There are also social media comments and captions that may offer some clues, or (if they haven't been cropped out) watermarks left behind by an AI generator in the bottom right corner of the image itself.
Importantly, an image may also contain glitches that give the game away, with the fact checkers noting that — for the time being, at least: "These defects are the best way to recognise a fabricated image".
Telltale signs include hands with the wrong number of fingers; asymmetrical ears; teeth and hair featuring odd outlines or textures; or other strange goings-on in the background (for example, blurring, distortion or — as one example revealed — missing legs).
AI can also struggle to generate accurate reflections, with one expert noting: "A good way to spot an AI is to look for shadows, mirrors, water, but also to zoom in on the eyes, and analyse the pupils since there is normally a reflection when you take a photo.
"We can also often notice that the eyes are not the same size, sometimes with different colours."
This is by no means an exhaustive list of things to look out for, and while tools such as the Hugging Face AI image detector may help, the fact checkers stress that common sense is key.
It's something to keep in mind if you find yourself wondering whether Russian President Vladimir Putin really was a buddhist monk or French President Emmanuel Macron was actually caught collecting rubbish on the street.
Fact checkers increased budgets and diversified content in 2022, report finds
Fact-checking organisations have upped their budgets and expanded their reach as misinformation grows more prevalent, a report from the International Fact-Checking Network (IFCN) has found.
The latest State of the Fact Checkers Report, released last month, found that the proportion of fact checkers operating with a budget of more than $US500,000 had grown from 19 per cent in 2021 to 25 per cent in 2022.
The share of organisations operating on $US20,000 or less fell significantly, from 26 per cent to 10 per cent of the 93 fact-checking outlets surveyed.
Ferdi Özsoy, the interim director of the IFCN, told the Poynter Institute that the increase in budgets demonstrated "the fact-checking community's steadfast dedication to preserving truth and accuracy amid the growing prevalence of misinformation in today's world".
The report also found that fact checkers were "moving beyond" initial areas of specialisation to "tackle misinformation across various public interest topics".
More than 95 per cent of the organisations surveyed said they debunked health misinformation and addressed political and social issues, with more than 88 per cent also checking economic claims.
It's not all good news, however, with the report also finding that fact checkers were increasingly reliant on funding from a single source: Meta's third-party fact-checking program.
According to the report, almost half (45 per cent) of all income for fact checkers comes via Meta's program, whereby accredited fact checkers (including RMIT FactLab, publisher of this newsletter) are paid to debunk misinformation on Facebook and Instagram.
"The financial landscape … remains uncertain, and organisations should proactively diversify their funding sources," the report warned.
"Increasing individual donations and membership subscriptions can reduce reliance on big tech grants and foster community-driven support."
Edited by Ellen McCutchan and David Campbell
Got a fact that needs checking? Tweet us @ABCFactCheck or send us an email at factcheck@rmit.edu.au
