Vodafone is to introduce a tough new global policy to prevent its advertising from appearing on fake news and hate speech sites.

The telecoms company, which spends close to £400m of its £750m annual global ad budget on digital ads, is thought to be the first to move beyond relying on Google and Facebook’s automated technology.

Working with Google, Facebook and ad agencies, it is setting up a “whitelist” of sites on which its ads can run.

“It is relatively easy to blacklist certain sites, such as those relating to porn and gambling, but it is difficult to exclude content that is – in editorial terms – at odds with an advertiser’s own principles and beliefs,” said Matt Peacock, group director of corporate affairs at Vodafone.

“When it comes to making a judgment about a site or channel focused on fake news or hate speech, what’s needed is a human judgment, an editorial assessment. It’s not possible to rely on algorithms alone. For that reason, we’ve concluded that the best way to ensure brand safety is to use a whitelist approach.”

Vodafone’s whitelist will feature sites it has identified as “highly unlikely to be focused on harmful content”.

“Hate speech and fake news threaten to undermine the principles of respect and trust that bind communities together,” said the Vodafone chief executive, Vittorio Colao. “Vodafone has a strong commitment to diversity and inclusion; we also greatly value the integrity of the democratic processes and institutions that are often the targets of purveyors of fake news. We will not tolerate our brand being associated with this kind of abusive and damaging content.”

The whitelist will be reviewed regularly to make sure it is “appropriate and neither too broad nor too narrow”.

Vodafone’s definition of sites peddling hate speech has been updated to include those that “deliberately intend to degrade women or vulnerable minorities”.

Its definition of fake news sites includes those “presented as fact-based news (as opposed to satire or opinion) that has no credible primary source (or relies on fraudulent attribution to a primary source) with what a reasonable person would conclude is the deliberate intention to mislead”.

The company said an outlet will only be blocked from its whitelist if its “predominant purpose” is hate speech or fake news, meaning it will not move to block tabloid or extreme rightwing or leftwing publishers if they occasionally are at odds with its new rules.

Vodafone stressed that its rules do not cover the type of content its customers can access.

Facebook, which has toughened its own policies on fake news on its site, said Vodafone’s new policy would work hand in hand with its own efforts.

“We welcome Vodafone’s new global rules,” said a Facebook spokesman. “We work closely with all of our advertiser partners to ensure that they have control over where their ads appear on our platform, and vigorously enforce our community standards which include policies against hate speech, violence or extremist content.”

Both Google and Facebook have moved to introduce tools to stem the spread of false information and minimise the chances of advertising appearing next to inappropriate content.

However, there is evidence that these are not 100% effective.

“While automated advertising is a powerful tool in a small minority of instances it can also lead to unintended and potentially harmful outcomes including advertising appearing next to offensive content,” said Vodafone. “Additionally, automated advertising technologies can have the effect of generating revenue for outlets focused on hate speech and fake news.”

Facebook said it would “continue to work with all advertisers to help them deliver great advertising experiences that drive business results”.

Google did not respond to a request for comment.