Social media, including Facebook, must proactively block content inciting hatred and prevent online campaigns which target minorities, such as those undertaken in Myanmar, the United Nations human rights chief said on Wednesday.

Zeid Ra’ad al-Hussein, U.N. High Commissioner for Human Rights, was speaking after U.N. experts accused Myanmar generals of “genocidal intent” and said Facebook had allowed its platform to be used to incite violence against Rohingya.

Facebook said on Monday it was removing several Myanmar military officials from the social media website and an Instagram account to prevent the spread of “hate and misinformation” after reviewing the content.

Zeid, whose spokesman said he has met with major tech companies in Silicon Valley, including Facebook and Google, in recent months, was speaking to a news conference before his four-year term ends on Aug. 31.

Zeid said he didn’t feel Facebook took the issue seriously at first but that the company’s attitude began to change after Yanghee Lee, U.N. special rapporteur on human rights in Myanmar, told a Geneva press conference in March that Facebook was being used in the country to spread hate speech.

“But it shouldn’t be because the press or the human rights community highlights the problem for them then suddenly to respond. They should be aware of it ahead of time,” he said.

“So I don’t think they should wait until the crisis begins. They should be thinking proactively about what steps they will take to mitigate that,” he added.

Facebook said on Monday that while it was too slow to act in the case of Myanmar, it was now making progress, with better technology to identify hate speech and improved reporting tools.

However, Zeid said there was a danger that social media could be over-regulated in a way that breaches human rights law including the right to freedom of expression.

Tech giants should “keep the broadest space available and open to the exercise of freedom of expression”, relying on international human rights law for regulation, he said.

U.S. President Donald Trump on Tuesday accused Google’s search engine of promoting negative news articles and hiding “fair media” coverage of him, vowing to address the situation without providing evidence or giving details of action he might take.

Trump’s attack against the Alphabet Inc. unit follows a string of grievances against technology companies, including Twitter Inc and Facebook, which he has accused of silencing conservative voices.

Facebook Inc., Twitter Inc. and Alphabet Inc. collectively removed hundreds of accounts tied to an alleged Iranian propaganda operation on Tuesday, while Facebook took down a second campaign it said was linked to Russia.

Facebook CEO Mark Zuckerberg said the accounts identified on his company’s platform were part of two separate campaigns, the first from Iran with some ties to state-owned media, the second linked to sources which Washington has previously named as Russian military intelligence services.

Officials in Iran, where it is a holiday to mark the Muslim Eid al-Adha festival, were not immediately available to comment. Moscow has repeatedly denied using hacking or fake social media accounts to influence foreign elections. The Russian embassy in Washington did not immediately respond to a request for comment.

The move by Facebook and others is the latest attempt by global social media giants to guard against political interference on their platforms. It comes as concerns are rising about foreign attempts to disrupt the U.S. midterm elections in November.

The United States earlier this year indicted 13 Russians for alleged attempts to meddle in U.S. politics, but the latest alleged Iranian activity, exposed by cybersecurity firm FireEye Inc. suggests the problem may be more widespread.

“It really shows it’s not just Russia that engages in this type of activity,” Lee Foster, an information operations analyst with FireEye, told Reuters.

FireEye said the Iranian campaign used a network of fake news websites and fraudulent social media personas spread across Facebook, Instagram, Twitter, Google Plus and YouTube, to push narratives in line with Tehran’s interests.

The Iranian mission to the United Nations did not respond to a request for comment.

The activity was aimed at users in the United States, Britain, Latin America and Middle East up through this month, FireEye said, and included “anti-Saudi, anti-Israeli, and pro-Palestinian themes” as well as advocacy of policies favorable to Iran such as the U.S.-Iran nuclear deal.

FireEye said the Iranian activity did not appear “dedicated” to influencing the upcoming election, though some of the posts aimed at U.S. users did adopt “left-leaning identities” and took stances against President Donald Trump.

That activity “could suggest a more active attempt to influence domestic U.S. political discourse” is forthcoming, Foster said, but “we just haven’t seen that yet.”

‘DISTINCT CAMPAIGNS’

Facebook said the Russia-linked accounts it removed were engaged in “inauthentic behavior” related to politics in Syria and Ukraine. It said that activity did not appear to be linked to the Iranian campaign.

“These were distinct campaigns and we have not identified any link or coordination between them. However, they used similar tactics by creating networks of accounts to mislead others about who they were and what they were doing,” the company said in a statement.

Facebook last month removed 32 pages and accounts tied to another misinformation campaign without describing its origins, but which U.S. lawmakers said likely had Russian involvement.

Microsoft said this week that hackers linked to the Russian government sought to steal email login credentials from U.S. politicians and think tanks, allegations the Russian foreign ministry described as a “witch-hunt.”

FireEye said the U.S.-focused Iranian activity ramped up last year, just months after Trump took office, with websites and social media accounts posting memes and articles, some of which were apparently copied from legitimate U.S. and Iranian news outlets.

In some cases, the domains for the fake websites like “US Journal” and “Liberty Free Press” were originally registered years before the 2016 election, in 2014 and 2013, but most remained inactive until last year, FireEye said.

Arabic-language, Middle East-focused websites appear to be part of the same campaign, the company added.

The technology companies variously said they linked the accounts to Iran based on user phone numbers, email addresses, website registration records and the timing of account activity matching Iranian business hours.

FireEye expressed “moderate confidence” about the Iranian origins, but said it has not been able to tie the accounts back to a specific organization or individuals.

Hundreds of thousands of people followed one or more of the Facebook pages implicated in the campaign, Facebook said.

It shared examples of removed posts, including a cartoon depicting an Israeli soldier executing a Palestinian and a fake movie poster showing President Trump embracing North Korean leader Kim Jong-un.

Postings cited by FireEye expressed praise for U.S. politicians and other Twitter users who criticized the Trump administration’s decision in May to abandon the Iranian nuclear pact, under which Iran had agreed to curb its nuclear weapons program in exchange for loosening of sanctions.

Some Twitter and Facebook accounts were designed to appear as if they were real people in the U.S., Britain and Canada, according to FireEye. The accounts used a combination of different hashtags to engage in U.S. culture, including “#lockhimup,” “#impeachtrump” and “notmypresident.”

Twitter, which called the effort “coordinated manipulation,” said it removed 284 accounts.

Facebook said it removed 254 pages and 392 accounts across its flagship platform as well as its Instagram service. Some of the accounts had events and groups associated with them.

The accounts spent about $12,000 to advertise through Facebook and Instagram using a variety of currencies, Facebook said. The company said it had notified the U.S. Treasury and State departments of the purchases, which may potentially violate sanctions.

Alphabet, which includes Google and YouTube, did not respond to a request to comment.

India has asked its telecom operators to find ways of blocking applications such as Facebook and messaging app WhatsApp in the case of misuse, according to a document seen by Reuters.

India has in recent months intensified efforts to crack down on mass message forwards after it found that people were using social media and messaging apps to spread rumours and stoke public anger.

WhatsApp in particular has faced the wrath of Indian regulators after false messages circulated on the messaging platform led to a series of lynchings and mob beatings across the country.

The department of telecommunications in July asked Indian telecom service providers, as well as mobile and internet industry bodies, to “explore various possible options” to block such apps.

“You are…requested to explore various possible options and confirm how the Instagram/Facebook/Whatsapp/Telegram and such other mobile apps can be blocked on internet,” according to the government letter dated July 18 and seen by Reuters.

Facebook Inc, which owns both WhatsApp and photo-sharing platform Instagram, declined to comment. Telegram did not immediately respond to a request for comment.

A source at India’s department of telecommunication said the letter was aimed at finding ways to block such apps during “emergency situations”.

“There is a need for a reasonable good solution to protect national security,” said the official, who declined to be named.

For WhatsApp, India is its biggest market with more than 200 million users and one where it says people forward more messages, photographs and videos than any other country.

Following calls from the government to stem the platform’s misuse, WhatsApp has moved to deter mass message forwards and launched an advertising campaign to educate consumers.

In July, WhatsApp said message forwards will be limited to five chats at a time, whether among individuals or groups, and said it will remove the quick forward button placed next to media messages.

Separately, India’s federal police has begun probing Cambridge Analytica’s misuse of Facebook user data, which New Delhi suspects included information on Indian users.

The Times has just completed a detailed investigation into the way that advertising and pay per view clicks fund extremist and other content on the Internet and in the last few days they have also highlighted how child pornography and other activity, including the promotion of extremist material, has taken place inadvertently through the algorithm systems set up by social media companies.

We complained to a social media company about 8 weeks ago about how their algorithms were promoting far right and Islamist material to members of Muslim communities in the United Kingdom. In particular, Islamist material was being promoted to individuals who had an interest in Islam, Muslims and facets related to both and the algorithms were inadvertently strengthening such groups who promote a world view of political Islam at the expense of the rights of others communities and with narratives that include a ‘war being waged on Islam’ and where Muslims are the ‘victims’ of the West. Some of these groups have also wavered on what is terrorism and when it takes place in our country.

We would urge social media companies to ensure that their algorithms do not support groups who are against universal values of human rights, pluralism, free speech, democratic values and the right to question and dissent in communities. These are fundamental values and the last thing we want is for society to be shaped by a set of algorithms who know nothing about the context and nature of groups.

We will continue to make the case on such issues and we will continue to move forward constructively on such matters with social media companies.