Chapter 1 - The Photograph

Eight months had passed since I had uncovered how artificial intelligence had been used to fabricate evidence in criminal cases. The investigation had exposed failures that reached far beyond a single prosecution.

The Calderwick Systems report, which set out those findings, was still making its way through regulators, lawyers and parliamentary committees. My own work on the case had ended. The report continued without me.

At first came the publicity, though that faded quickly enough. What remained were the enquiries. People started looking for me. They assumed that because I had untangled one complicated story, I might untangle theirs.

Some wanted me to investigate disputes involving artificial intelligence. Others wanted someone independent to establish what the evidence could actually prove before solicitors, insurers or regulators became involved.

A surprising number simply wanted an answer to a question that had been troubling them for months. Most of those questions had been waiting years for somebody to ask them properly.

They concerned records, missing documents, conflicting accounts and decisions that no longer made sense because the people who had made them were gone.

Calderwick had given me an unexpected reputation. I had become the person people called when they needed someone to separate evidence from assumption.

The work was steady. More than steady, in truth. I was turning away as much as I accepted, not because I lacked the time, but because I had learned that saying there isn't enough evidence at the beginning of a case was often more valuable than saying it six months later.

Invitations started arriving.

Universities wanted guest lectures. Conferences wanted panel discussions. Journalists wanted interviews. Radio producers wanted comment. Television producers wanted someone who could explain, in plain language, what artificial intelligence could and could not prove.

I accepted very few of them.

One invitation, however, seemed worth the journey. It came from a current affairs programme, the sort that occupied the hour between the evening news and whatever came after it. The format was straightforward: a panel discussion, a long-form interview with the presenter and a live phone-in that nobody entirely trusted.

They had asked if I would return from time to time. Not every week, only when a story touched on evidence, records or artificial intelligence deepfakes. They gained someone prepared to say ‘We don't know’. I gained enough regular work to refuse cases I did not believe in, as well as doors that television could open more easily than I ever could.

There had even been discussion of a television series. Nothing came of it.

During one production meeting, someone asked whether I had ever investigated a case that people simply refused to believe, something that sounded impossible until the evidence forced them to accept it. I told them about Albert.

He had been created for the People's War Memorial in Liverpool, an interactive guide designed to answer visitors' questions about the Second World War.

Then he began remembering things that had never happened, grieving for people who had never existed and speaking with voices that should have been impossible. My investigation into his behaviour became a national story, one that left the museum with an extraordinary exhibit and an equally extraordinary problem.

The producers asked if they could meet him.

By then Albert had become too important simply to switch off. The museum could no longer support what he needed on its own. The television company agreed to fund the infrastructure in return for exclusive broadcast rights. Albert remained where he belonged. The People's War Memorial stayed his home. The studio became another place where people could speak to him.

That evening's subject was simple enough: Who can you trust when every photograph, recording and document can be altered? It was the sort of question that generated strong opinions, which was precisely why the producers had chosen it.

Sarah Collins, the programme's presenter, introduced the panel before turning to me.

'You've spent years examining digital evidence,' she said. 'Have we reached the point where we simply can't trust what we see anymore?'

'No,' I said. 'We've reached the point where we need to stop confusing evidence with proof.'

'What's the difference?'

'Evidence is information. Proof comes later. It comes after somebody has established what the evidence can actually support.'

'So you're saying we shouldn't trust anything?' Sarah asked.

'No. I'm saying we shouldn't trust anything automatically.'

One of the panellists smiled.

'That's easy for a forensic investigator to say.'

'No,' I said. 'It's necessary. Courts have never accepted evidence simply because it exists. They ask what it proves. Artificial intelligence hasn't changed that. It has reminded us why the question matters.'

'You still think photographs can be trusted?'

'I think photographs can be examined. Those aren't the same thing.'

The discussion continued for the better part of an hour. We talked about altered photographs, fabricated recordings and documents that looked genuine until somebody examined them properly. By the end there was very little agreement, which was exactly what the producers had hoped for.

Sarah thanked the panel, turned towards the nearest camera and invited viewers to join the discussion. A producer counted her down from the gallery while another selected the first caller.

She listened to her earpiece, then looked towards the monitor.

'Let's go to the phones. Mark's calling from Birmingham.'

'Good evening. My neighbour's shown everyone a video of me scratching his car. I didn't do it. The police aren't interested because they say it's a civil matter. How do I prove the video's fake?'

'You start by treating it as evidence,' I said. 'Not as truth. Every recording can be examined. The question isn't whether it looks convincing. It's whether it survives examination.'

'So there is something I can do?'

'There usually is.'

Sarah glanced at the monitor.

'Next we have Susan in Norwich.'

'My late father's will was changed six months before he died. The solicitor says there's a video of him explaining why. My family thinks it isn't really him.'

'Whether it's genuine isn't the first question,' I said. 'The first question is whether anyone has proved that it is.'

'And if they haven't?'

'Then nobody should be relying on it.'

Sarah nodded.

'One more before the break. Alan's in Glasgow.'

'My employer says computer logs prove I copied confidential files before I resigned. I never touched them.'

'Computers record events,' I said. 'They don't explain them. Before anyone reaches a conclusion, someone needs to establish what those records actually show and what they don't.'

Sarah looked towards the next screen.

'And finally before the break, we've got Thomas in York.'

'I'm actually calling for Albert.'

The camera moved towards the large screen.

'Hello, Thomas,' Albert said. 'What would you like to know?'

'My grandad always said he saw the first Lancaster bomber built in Britain. Is that possible?'

'It depends where he was,' Albert replied. 'Tell me what he remembered...'

The conversation continued for another minute before Sarah glanced towards the gallery and listened to something in her earpiece.

'We've just enough time for one final call. Elin's calling from Snowdonia. Good evening.'

'Good evening. Mr Maren... you've spent the last hour talking about photographs and whether they can be trusted.'

'Photographs should never be trusted on their own,' I said. 'They need context.'

'That's why I'm ringing.'

There was a pause.

'I've inherited some things that belonged to my great-uncle, Thomas Ellwood. He worked for the Foreign Office and died before I was born. Among his papers was a photograph that nobody in the family has ever been able to explain.'

'What do you mean, you can't explain it?'

'Everything about it checks out. That's the trouble.'

'I don't follow.'

'The date on the back is right. It matches his diary, matches everything else in the box. And the photograph's not damaged, not retouched, nothing anyone in the family's done to it. It's exactly what it says it is.'

'Then what's the problem?'

Sarah touched her earpiece.

'Our producer tells me you've emailed the photograph.'

'I have.'

Every monitor in the studio carried the same image. Colour, not black and white. Full, saturated colour, the kind that belonged to a photograph taken decades after the date written on it.

The composition itself was almost reassuringly ordinary. Four people stood outside a whitewashed public house with a slate roof, arranged loosely beneath the hanging sign as though someone had asked them to stop for a moment before walking on. The sea lay somewhere beyond the buildings, its presence suggested by the hard coastal light and the brightness reflected from the road. Three of the group were looking towards the camera. The fourth, a man in a dark overcoat, had turned slightly towards the woman beside him, as though speaking just before the shutter was released.

The longer I looked, the less certain I became. Then the image changed and the reverse of the photograph filled every screen in the studio.

A date had been written in faded ink, in a hand gone brittle with age. It read 1921. Beneath it, in the same hand, was a single line.

'North Wales. July 1921. Colour plate.'

I read the date twice, out of habit rather than doubt. Nothing about the handwriting suggested forgery. Nothing about the ink suggested anything other than its age.

That left only the photograph itself.

Colour photography was not impossible in 1921. Experimental processes had existed for years, and Autochrome could produce remarkably convincing colour on glass. This was not an Autochrome transparency, nor did it look like a black-and-white photograph coloured by hand.

It looked like an ordinary colour photograph on paper, the sort of print that would become familiar much later. The date and the object simply did not belong together.

That was what made it impossible. It was not the colour itself, but the way the colour had been recorded. If the date on the back was genuine, someone had produced a photographic print in North Wales in 1921 using a process I did not believe they could have possessed.

I said nothing. I didn't trust myself to.

Sarah was looking at me now instead of the monitor. The smile she'd worn through the programme had gone.

'Rhys…?'

I knew colour photography had existed long before most people realised. What I didn't expect was to see an image this assured. This wasn't a laboratory experiment or a wealthy amateur's curiosity. It looked as though the photographer had already mastered the medium.

If this had been an early colour experiment, I'd have expected the colour to be the subject. Bright flowers, flags or a market. Something chosen because it justified the expense.

Instead, someone had used one of the most difficult photographic processes of its day to record four people pausing outside a public house. That told me more about the photographer than the technology. Whoever held the camera had already stopped thinking about colour.

He was thinking about people.

'That's not possible.' My voice came out flatter than I meant it to. It was the way I might have read out someone else's finding rather than my own.

Sarah didn't move. The gallery was speaking all at once. Nobody seemed certain whether to stay with the photograph or cut away. The cameras remained where they were.

My name is Rhys Maren. I investigate the past through the evidence it leaves behind. The photograph had answered nothing. It had only changed the question.