The fallout from a U.K. IT scandal

Welcome to Runtime! Today: Why Fujitsu executives spent the week apologizing for their enterprise software in a U.K. courtroom, Microsoft got hacked, and the quote of the week.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

Dodgy in, dodgy out

Over the last several weeks, the U.K. has been rocked by the re-emergence of a long-running scandal involving over 700 people who were wrongfully terminated from their positions at the Post Office, which is some kind of combination FedEx/Bank of America government-run company that is apparently allowed to prosecute its employees. Those employees were convicted of stealing money from the company after its Horizon IT system — built by Fujitsu — falsely claimed that funds were missing from accounts they oversaw.

Employees claimed for years that the Post Office and Fujitsu were aware of serious flaws in Horizon but allowed the prosecutions to go forward anyway. Much of the story had already been reported, but the case exploded into the public consciousness in early January after a four-part television series showcased some of their stories, and led to Fujitsu's European leader, Paul Patterson, apologizing for his company's actions in court Friday.

• "To the subpostmasters and their families, we apologize, Fujitsu apologizes and is sorry for our part in this appalling miscarriage of justice," Patterson said Friday, according to The Guardian.
• Fujitsu not only knew that Horizon was flawed pretty much right after it was installed in 1999, but Fujitsu also "had a 'don’t share with the Post Office' approach to a document chronicling the known errors in the system," The Guardian reported.
• For years Post Office employees had "complained about bugs in the system after it falsely reported shortfalls — often for many thousands of pounds," according to the BBC.
• However, Post Office management ignored those complaints and worked with Fujitsu to discredit their claims, which led to hundreds of convictions, financial ruin, and prison time.

The Post Office scandal is a cautionary tale for vendors and buyers alike as enterprise tech plunges headlong into generative AI as well as enterprise software that promises "a single source of truth." 

• Even before the launch of ChatGPT, software companies were marketing AI tools that could detect the emotions of people on sales calls, evaluate job applicants, and make lending decisions.
• Flaws are pretty much inevitable in software, even if vendors are reluctant to admit that fact.
• But when that software is being used to alter the course of someone's life, those mistakes can have outsized consequences, and it's not clear how many people in a rush to build generative AI software are considering the risks involved.
• After all, people tend to trust what the computer tells them.

Still, it's also clear that government agencies should be held to a higher standard when implementing IT systems that affect so many people.

• One obvious lesson is that more diligence needs to be performed in the testing phase before such systems are fully deployed to production.
• But given how closely it worked with the Post Office to downplay the flaws in its software, Fujitsu is likely going to have to cough up a rather large settlement to compensate the victims of the miscarriage of justice.
• Custom government IT projects are tricky things to get right, which underscores yet again the need for the public sector to attract the kind of tech talent that wouldn't have let these kinds of problems spiral out of control.
• Given how fast the AI boom is expanding, they might not want to wait much longer.

Once more unto the breach

Microsoft notified the SEC Friday that the hacking group believed to be behind the SolarWinds attack broke into email accounts belonging to unidentified "members of our senior leadership team" and stole both emails and documents from those accounts.

The incident took place in late November but wasn't discovered until last week, according to a company blog post. The attackers "used a password spray attack to compromise a legacy non-production test tenant account and gain a foothold" before figuring out how to get into the executives' accounts as well as accounts belonging to "employees in our cybersecurity, legal, and other functions," the company said.

Needless to say, there was likely some extremely valuable information present in those email accounts when they were breached. The good news is that no customer accounts were affected, and Microsoft said it will "act immediately to apply our current security standards to Microsoft-owned legacy systems and internal business processes, even when these changes might cause disruption to existing business processes," referring to its pledge last November to double down on security.

Quote of the week

“Every one step we take closer to very powerful AI, everybody’s character gets like +10 crazy points. It’s a very stressful thing — and it should be, because we’re trying to be responsible about very high stakes.” — OpenAI co-founder and CEO Sam Altman, describing the upheaval at this company last November and giving us all something to look forward to.

The Runtime roundup

Meanwhile, Altman wants to raise billions to launch a network of chip-making factories, according to Bloomberg, which would involve a non-zero amount of crazy points.

AWS will spend $15 billion in Japan to expand its capacity over the next three years, adding to the two regions it currently operates in the country.

Supermicro shares hit an all-time high Friday after it released blowout preliminary revenue numbers for the fourth quarter, signaling a surge in the server market.

IBM acquired "application modernization capabilities" from Advanced, a U.K. company that specializes in mainframe applications.

Oracle plans to build a data center in Israel that will operate nine stories underground, according to CEO Safra Catz.

Thanks for reading — see you Tuesday!