Your digital past is now part of the job interview

Artificial intelligence is allowing employers to examine years of online activity within hours, turning forgotten posts, anonymous accounts and even deleted profiles into potential career liabilities. The warning that jobseekers should watch what they publish online is hardly new. What has changed is the speed, depth and persistence with which companies can now search. Employers that once relied on a quick Google search are increasingly using artificial intelligence to connect information scattered across social networks, news reports, photographs, usernames and other public records. In some cases, that scrutiny continues long after the employee has been hired. The expanding practice was emphasised by Callum Borchers in The Wall Street Journal, which reported that businesses are digging further into candidates’ internet histories as they place greater weight on character, reputation and cultural fit. An old Facebook post, a forgotten Reddit account or a reused username can now become part of an unofficial employment file. Screening company Ferretly, whose reported clients include Deloitte, Ally Financial and advertising agency BBDO, uses artificial intelligence to analyse online behaviour that conventional background checks may overlook. Its technology can examine material across several platforms, including text, photographs and memes, looking for content connected to harassment, violence, prejudice, drugs or other reputational risks. According to the Journal, such detailed screening was once largely reserved for executives and other senior figures whose behaviour could embarrass a company publicly. However, cheaper and faster technology has extended it to customer-facing positions and, in some businesses, much further down the organisation. The industry’s scale is already considerable. Rival screening company Fama says it has delivered more than 45 million reports to over 3,600 customers, examining public online activity for signs of workplace-related misconduct. Its services include pre-employment checks as well as continued screening of existing staff. Consequently, the background check may no longer end when a contract is signed. Some employers are paying for continuous monitoring, hoping to identify threatening, discriminatory or otherwise damaging behaviour before it develops into a workplace or public-relations crisis. However, the technology can reach beyond posts published under a person’s real name. Accounts may be linked through repeated usernames, photographs, biographical details or digital identifiers used elsewhere online. Someone using an alias on an adult-content platform or prediction market may still reveal their identity by recycling a social-media handle or connecting activity to a traceable cryptocurrency wallet. The Journal reported that Ferretly generally alerts an employer when it is at least 70 per cent confident that particular online material belongs to the person being screened. What happens next is left to the employer. That threshold also exposes one of the system’s greatest weaknesses. A probability is not proof, while humour, irony, political debate and material quoted for criticism can be difficult for automated tools to interpret. Similar names, copied photographs and impersonation accounts can further increase the danger of an innocent person being linked to somebody else’s behaviour. There are legal limits, too. In the United States, employers using specialist screening companies must comply with the Fair Credit Reporting Act. Federal guidance requires them to notify candidates, obtain permission and provide an opportunity to challenge a report before taking adverse action. Employers must also ensure that background information is not used in a way that discriminates on grounds including race, religion, sex, age or disability. European employers face additional restrictions under data-protection law. Britain’s Information Commissioner’s Office warns that intrusive checks using specialist software are unlikely to be lawful, appropriate or necessary, particularly when they reach information that is not publicly available. Meanwhile, the EU AI Act treats systems used to analyse applications, evaluate candidates or manage workers as potentially high-risk. Such tools face requirements concerning transparency, data quality, traceability and meaningful human oversight. For workers, the obvious response may be to delete old accounts, tighten privacy settings and remove embarrassing posts. Those measures can reduce exposure, although they cannot erase screenshots, archives or copies stored elsewhere. Even a perfectly empty digital history may create a different problem. Some verification businesses are developing tools designed to flag applicants whose online presence appears unusually shallow or newly constructed, partly because fabricated identities and fraudulent job applications have become easier to create. Vinda Souza of applicant-verification company RefAssured told the Journal that when it becomes clear someone once had an online presence and suddenly does not, “as a hiring manager, that would invite inquiry”. The safer approach may therefore be neither oversharing nor disappearing. Candidates can search their own names, review abandoned accounts, correct inaccurate information and build a credible professional presence that gives employers reliable material to find. For years, people were warned that the internet never forgets. Artificial intelligence has added a more unsettling reality, it can now remember on an employer’s behalf.