Hacker News new | comments | show | ask | jobs | submitlogin
Amazon Execs Addressed Concerns About Rekognition/ICE at All-Hands Meeting (www.buzzfeednews.com)
65 points by minimaxir 4 months ago | hide | past | web | 101 comments | favorite

I watched the all hands today. I just want to comment on the journalism here. For one, whoever this employee is that email buzz feed news is taken as representative to general feelings for all of Amazon employees which is ridiculous. Also the article makes this assertion "Though the company-wide Q&A was not meant to only address concerns about Amazon’s facial recognition deals with law enforcement, it was a major topic on employees’ minds.", which it has no data to back up. One question at the tail end of a 90 minute all hand does not make it a major topic.

I think you make a solid point here. I do think Amazon and its employees should be debating their concerns on this topic. But if the only question/talk of Rekognition vis-a-vis law enforcement during the all-hands is that one question quoted by the article, then that doesn't justify a news story just on that one question at the meeting. That's not to say BuzzFeed isn't justified in continuing to report on the controversy and debate, just that this recent moment at the all-hands isn't news in itself.

Before the big Google walkout this week, there was at least one HN discussion about it a few days prior, with some commenters questioning the purpose/courage of a walkout if the employees aren't willing to quit. Well, even if Google employees didn't put money where their mouths were, participating in the walkout at the very least is a public statement. And when 20,000 [0] employees make that statement, there is no debate at all if a reporter describes the Google controversy as one that has sparked significant company-wide strife.

looks like you have an ex-Googler working at your company.

News should only be taken with a grain of salt in 2018

More like, news from certain sources should be taken with a grain of salt.

There are plenty of good journalists with high reputation that you can certainly take more than a pinch of salt with.

Even reputations are hotly debated in this climate. I think a good habit is to read from as many sources as possible, including controversial ones, and making a private determination as to how to weight reporting.

There are enough sources supporting any worldview that there's a real risk that you'll just end up supporting whichever one you started with. I have no idea how you can do better though.

I struggle with confirmation bias, myself.


Anything that hits the front page of nytimes.com or washingtonpost.com for example. Opinion pieces are a different ballgame, obviously.

There's no shortage of people who share that same "belief", but would end up writing down a different and conflicting list of people than you, with no loss of accuracy

This is not some new phenomena that didn't exist until very recently. Critical thinking has been required since, well, 'news' was a thing.

I agree it’s not a new phenomena and never said that it was. I’d argue that our news was historically fair-minded up until the recent mass-scale polarization.

So you are here to tell us that employees don't actually care as much as reported about the ethics implications of their work, and that leadership didn't actually meaningfully address the topic.

... is that meant to be a comment in defence of the company?

The comment is not really addressing the topic at hand, just how the article here makes bigger claims than it actually has data to prove.

Not addressing the topic at hand, preferring to discuss technical points is precisely the industry tend that concerns me.

You are perfectly welcome to find another industry if you don’t like this one.

Please don't get personal in HN arguments.


This is a disingenuous statement. Given the common understanding that "software is eating the world," your suggestion is about as earnest as "you are perfectly welcome to find another planet."

This is Hacker News, not Manager News or Lawyer News.

> trend that concerns me.

No one cares about what trends concern any one person.

> we feel really great and really strongly about the value that Amazon Rekognition is providing our customers of all sizes and all types of industries in law enforcement

Translation: "We are making so much money that we have entirely sold our ethics and will continue doing whatever makes us more money."

They haven't sold their ethics because they didn't have those ethics in the first place, and haven't ever seriously tried to pretend otherwise. Amazon is pretty transparently in business to make money, with little regard for anything else. The closest thing they have to ethics is "Customer Obsession", but even that is explicitly justified as a means to maximize long-term profit.

I really don't know how this response surprised any of their employees (as a previous employee, it certainly does not surprise me). Those who can't abide it oughtn't work there; this isn't something that is going to be changed from the inside.

I mean, yeah. Amazon's been doing that for a long time.

Back in the day, Amazon was very fond of repeating over and over that customer trust and customer loyalty was the very most important thing. Customers absolutely had to know that if they ordered something on Amazon, the right product would arrive on time. A data leak would be seen as an existential crisis. Shipping the wrong product was a customer service emergency.

Then Amazon found out how much money it could make from being a third party marketplace and mostly just giving up on quality control. Now nobody has nearly the confidence that the thing they're buying is a real thing. Is that USB-C cable gonna work with your phone? Who knows? Amazon doesn't. But they'll be happy to sell it to you.

Turns out that whole "customer first" thing was more of a temporary tactic more than it was any sort of moral grounding. It was a great tactic. It got Amazon to #1 from an era where nobody trusted eCommerce. But they dropped it once they didn't need it.

Remember, Amazon's the place that kicked a political organization off of their cloud with no notice because Senator Lieberman placed a quick phone call to Jeff Bezos and suggested it'd help him win GovCloud if he did.

This generalization is a bit over dramatic. Not everyone has problems just because hackernews has articles about mixed lots and other quality concerns. If it was as bad as you say they would not be pulling in the amount of revenue they are.

What is it about executive-speak? I entered the Google announcement about changes in their sexual harassment with a bad feeling because it started with a very executive-sounding statement that actually left me surprised with what followed.

If language creates this negative reaction, isn't that the opposite of the desired result? Or am I just particularly skeptical?

Executive speak is designed for minimal fallout. But it has other deeper value in maintaining power structures and internal org stability.

Any deviations and rumblings from within the hierarchy become very easy to identify if they don't use the same language.

Disagree. The technology is out there, if its not amazon, its going to be someone else.

I agree with Andy. It's not up to the company to protect civil rights through limiting the use of their product -- that's the job of the government. If Amazon stopped allowing law enforcement to use their products, someone else would literally just resell their products, because there is no way you can enforce every use, and you've just kicked the can down the road.

If Amazon really wanted to take a stand on this, they would use their massive size and lobbying power to lobby the government to put in new rules and laws about the use of facial recognition technology.

Not only would that help everyone at once, but it would also protect their profit margins because then every company has to play by the same rules, thus allowing them to justify the expense to their shareholders. It's a win all around.

It's the responsibility of all of us to protect civil rights.

Like I said in my post, I think they should, but not through banning use of their product. They should be protecting civil rights through lobbying.

So you think they should protect civil rights through lobbying while simultaneously profiting off of the erosion of civil rights? What about leading by example?

Telling Amazon to lead by example is about as fair as telling Warren Buffet to "just pay more taxes then" when he complains that rich people don't pay enough taxes. It doesn't work when only one player does it. That's why Amazon should lobby, and that's why Buffet spends his time lobbying while simultaneously employing accountants to reduce his tax bill.

Why is the argument that they must profit off it while legal while also lobbying against it? Why can't they lobby against it and not profit from the erosion of our civil rights?

Kant would like a word with you about moral imperatives.

If you think something is unethical, then facilitating it for a profit is also unethical. If you don't think it's unethical, fine, but you don't get to have it both ways.

I don't think using facial recognition is inherently unethical. I think what you do with it may be unethical. That's why government regulations would be much more effective here.

What about the civil rights of American citizens who don't want their votes and social services diluted by illegal immigrants unlawfully entering into the country?

No, illegal immigrants voting directly is not what I mean. But their children vote.

Just so happens the fastest demographic growth especially in the lower border states is the same as those which are immigrating illegally. They vote 70% for liberal policies and politicians. Is this fair to American citizens who may ideologically oppose these voters but can't illegally import their own votes to keep pace?

Are you suggesting that it’s not fair for individuals to vote who have a constitutional right to vote?

Instead of attacking the constitution I’d suggest critiquing the parties (and their policies) in power.

Why do you want to cite the legality of 2nd generation voters voting, but not cit the illegality of their parents coming here?

It's not fair to embrace law breaking when the consequences disenfranchise 1 party, 1 ideology, along w/ the pressure it applies to wages and social services demands.

I'm quite simply stating, it is unfair to turn your back on law breaking that may benefit some people (and of course certainly benefits the immigrant) while harming others, who happen to be citizens the law is designed to protect.

Can you articulate to me why you think we even have a law against illegal immigration, before you dismiss my point of view out of hand?

> it is unfair to turn your back on law breaking that may benefit some people (and of course certainly benefits the immigrant) while harming others

Who exactly is harmed by someone coming to this country? They don't qualify for any benefits paid for by tax dollars. They do however pay taxes. So in fact they are a net positive to anyone who is already here.

I'm curious how you think they harm a citizen just by being here.

The point you made above was that the illegal immigrants are "diluting your vote":

> What about the civil rights of American citizens who don't want their votes and social services diluted by illegal immigrants

This is a misrepresentation of the facts. Yes its true that these children would not have been American if their parents hadn't immigrated to America. However, they _were_ born in America and the immigration status of their parents has no legal baring on their citizenship or right to vote. They have as much of a legal right to vote as any other American.

> It's not up to the company to protect civil rights through limiting the use of their product -- that's the job of the government.

People and companies have a responsibility to behave ethically, whether the law specifies that or not.

Is there anything inherently unethical about facial recognition technology or does the concern lie in the potential for misuse and profiling by law enforcement?

AFAIK Rekognition just takes in a face and spits out an ID that you can use to identify an already known face. This doesn't seem inherently bad to me. Any misuse or profiling in this case is designed, managed, and implemented by law enforcement and the moral responsibility lies with those who create the policy, not a tool that they use to enforce it.

There's nothing inherently unethical (read: necessarily harmful) about building and selling facial recognition technology.

Nor is there anything inherently unethical (read: necessarily harmful) about manufacturing and selling opiates, narcotics, bombs, or guns.

The issue comes down to where along the product lifecycle you decide to end or limit the responsibility for how the product you build and design is used. And the mental calculations you do to weigh the positives/negatives of the technology you create and sell.

Obviously a manufacturer of bombs knows their product is going to be used to kill. If your code of ethics dictates that murder is bad, the calculation becomes - how "involved/complicit" am I in murder if I build the tools to murder, but don't use the tools myself. If we don't kill the bad guys, they might kill us. And hey, everyone in this society that pays taxes to fund these military and defense programs is complicit. I'm just doing what society needs to protect itself.

With facial recognition technology, if your code of ethics dictates that potential invasion of privacy/government tracking is bad, you would do a similar calculation. These people tracked are already in a database of "bad people" (aren't they? am I and my family in a database? do I have anything to hide?) - and society dictates we need this for our own protection, so what's the problem? If I make enough money selling these systems, I can buy my way to privacy, isn't that the way it should be?

Different people will come to different conclusions...there's no cut and dry answer, and infinite justifications for producing and selling any powerful technology/tool.

The spirit of my reply is essentially that you can't wait for the government to tell you what is or isn't ethical by enacting laws and regulations. A firm needs to be able to think for itself what kind of company it wants to be, not just blindly maximize shareholder value to the greatest extent the law will allow.

What is or isn't acceptable use of facial recognition tech is not an easy question to answer; I don't have an answer. Those making and using it have a duty to give it serious thought and adjust their actions accordingly.

First off, I disagree. Any scientist should be aware of what they are building.

Second, they actively market to law enforcement.

Fair enough if you disagree!

Regarding actively marketing to law enforcement, is there any law enforcement use case that is not unethical? For example, if a facial recognition system is set up to only alert on faces that are listed in the Interpol Most Wanted list is that system unethical?

It seems that there's a moral line between identifying faces and profiling/targeting people arbitrarily. It's impossible for Amazon to know how any given customer will use its products before they actually use them so the only reasonable ethical expectation is for Amazon to outline improper uses in its TOS and go from there.

EDIT: What if the above facial recognition system doesn't use automated software but instead relies on humans reviewing video feeds and trying to determine if a face matches one on the list? Does that change anything?

So marketing to law enforcement is unethical? What about all the database vendors that sell to law enforcement -- Oracle, Microsoft, or Postgres? Can't you see the nuances in that?

Law enforcement is the tool designed by the society to maintain the law. That inevitably lead to conflict with civil rights and abuse of power. A healthy democracy needs to protect human rights while giving the law enforcement the tools they need to effectively do their job. Constitution, legal framework, and freedom of expression are essential.

> So marketing to law enforcement is unethical?

Yes. The police state is unethical and doing anything to further it is unethical.

> People and companies have a responsibility to behave ethically, whether the law specifies that or not.

You are paying the salaries of the people you claim are acting unethically. That seems like more direct involvement than what amazon is doing.

Well now that's the real problem, they don't. They will face no real consequences for this. In fact they will probably fight hard to ensure they can keep making money, no matter the problems, right up until the day their own math says they need to flip the script.

Topf and Sons made ovens, it was the government that decided what to do with them.

Amazon deserves no cover on this: when the buyer uses your products for evil and you know this it is evil to continue to sell them. If they want to stand up at an all hands and make the argument that what ICE is doing isn't evil fine, that's their privilege, but they can't just wave their hands and say its not their problem. And if they did try to make that argument I'll bet a good portion of those 500,000 employees would be disgusted.

What about when one of your employees does something unethical and you continue to pay his salary, or have you gone on a tax strike?

> It's not up to the company to protect civil rights through limiting the use of their product -- that's the job of the government.

...and the employees, and the members of the board, and shareholders.

Regulatory capture of facial recognition technology might be a successful tactic - it certainly worked in Oracle's favor with government certified databases.

It's bizarre how many people pick ICE as their hill to die on.

It shouldn't be that surprising given the family separation fiasco.


I'd expect people so fired up about this issue to at least understand the basic facts.

Both of them are under "Homeland Security", and their 'mission' often overlaps, sooo perhaps it's more accurate to say 'homeland security.'

People with emphathy or ethics can get jumpy about ripping children away from their families. I wish I could still find it bizarre that some people don't.

Maybe because those people know that the alternative is to turn a blind eye to child trafficking. You can't both allow children to stay with whoever brings them and protect them from trafficking.

I mean CBP deserves way more shit than it's getting, but I don't really think hating ICE is that weird.

When you believe the bill of rights applies to everyone in the world which is becoming a common view, ICE seems like the perfect hill to die on.

I dunno, it kind of makes sense. Besides the IRS they probably have the most negative brand association in all of the US government. Their activities, especially recently, make them pretty easy to characterize as the bad guys. It’s not that crazy.

ICE is as good an example of the police state as anything else.

I for one think Andy's response is articulate and well explained. The technology does have a lot of potential to do good, and it seems that everyone is focusing on the bad side of it.

"Rekognition is actively been used to help stop human trafficking, to reunite missing kids with parents for educational applications, for security and multi-factor authentication to prevent theft."

Can it be used for evil? Yes. Should Amazon do its best to prevent it? Sure. But at the end of the day that is the responsibility of the government.

Who's the marketing genius that chose to spell it with a "k"? That just screams authoritarian dystopia to me.

"Let's call it 'Cloud Vision.' Oh, shit, somebody took that? Fuck, that was my big idea. Okay, no worries, we'll call it 'Image Recognition.' What, IBM has it AND it's a generic term? 'Cloud Sight'? No? Shoot, the meeting is in 10 minutes. Let's just go with Recognition, but, like, make it different. eRecognition? No, recognitionly? No. Re:cognition, like thinking! Yeah, now we're cooking! Josh, we're naming it Re:cognition!"

10 minutes later, Josh gets his kerning wrong in PhotoShop, and Re:cognition becomes rekognition.

I don't get why everyone is so riled about Amazon's service. At this stage, the tools to do it are widely available and have been simplified to the point that building a rudimentary facial recognition system is a novice programmer project.

It's too late to put this technology genie back in the bottle. Legislation would be our only salvation.

> “I don’t think Amazon leadership addressed the concerns brought up in the question, and I don’t think a Q&A session at an all-hands meeting is enough or suitable for addressing the concerns appropriately,” an Amazon employee, who requested anonymity, told BuzzFeed News in an email.

The response given by Jassy during the all-hands meeting doesn't really answer the underlying question, which is "how are you going to stop this technology from being abused". It starts off with "oh, but look at all the good it can do; of course we can't just stop" and then just drops the issue and says "the ToS will fix it".

I just searched for Rekognition and found it interesting that two days ago BI wrote this post:


> ... preparing to confront Jeff Bezos

Sounds like that went quite swimmingly.

> There is no way to hold leadership responsible in comparison to, say, a letter or email from HR to all employees where a statement is written and more concrete

Wouldn't the answer be to just leave your job? Don't support the company if you don't like what they are doing...

While I don't want to back up this terrible buzzfeed tabloid article, your argument holds absolutely no water. If the answer to all company injustices was to "just leave your job" why do we need any worker protections/wage regulations? In the real world many, many, many people don't have that option.

Working within systems to change behavior you don't like isn't an invalid approach. In the public sphere we often do it by protesting and voting, not by moving out of the state or country.

> He added that he thought it was the government’s responsibility to help specify regulations around the technology.

That's true but it doesn't absolve them of responsibility. They are choosing their actions willingly, and with full knowledge of the potential consequences.

This sort of behavior is inexcusable. It's just wrong.

What a non-issue. Snore...

If the intersectionalists had it their way, they'd ban Amazon from selling any and all services to all agencies and people whose politics they disagree with. They're yearning to stop serving Republicans and conservatives - almost 50% of the population.

That's OK, just as long as these intersectionalists are the first to go when revenue declines from their extremist philosophy stops supporting Amazon jobs. I think they should take their activism to it's logical conclusion and start cannibalizing both revenues and operations with their manufactured outrage and activism - only then will this reach fever pitch and force management to openly call this group's hypocrisy out.

We really need center of the aisle, common sense to be popular again in this country

"Intersectionalist" isn't synonymous with "leftist" or whatever you're trying to say. It's hard for me to see how intersectionality is relevant to the subject of Amazon's work with ICE.

It's VERY relevant - regular Amazonian employees do not care about selling to US law enforcement - the agencies that protect and allow Amazon to thrive and exist. This whole anti law enforcement narrative has been popularized by inter sectionalism - in whose heads law enforcement is full of majority race, privileged individuals oppressing the minorities.

> regular Amazonian employees do not care about selling to US law enforcement

Sorry but you don’t speak for regular Amazon employees. You’ve just made that up.

And what data can you offer to prove that a majority of Amazonians support not serving law enforcement?

You made the claim first mate.

Don't weasel out of supporting it by asking him to prove a negative as a response.

Current AMZN employee here. I can anecdotally say that I personally see no problem with selling the technology.

There was a extremely active discussion on an internal mailing list when "the letter" came out that was signed "Amazonians." While there were a small number of very vocal proponents of stopping the sale of the technology, the prevailing opinion was split between indifference and anger by less than 400, mostly young and relatively new Amazon employees, representing themselves as anywhere near a significant portion of "Amazonians" (however much that was intended versus sensationalism, I refuse to speculate).

I think you're speaking too broadly. I for one am very concerned with facial recognition, but it has nothing to do with immigration/left-right/minorities.

I think the government having too much power is more of a liability than it is a benefit.

I can agree with this wholeheartedly. But the opposition at Amazon is not what you stated. Anyways, I concur.

The logical conclusion of your arguments basically Fascism - putting profit above people, which isn’t remotely close to common sense. Nor is labelling those who oppose locking children in cages as “extremists”.

Shouldn’t businesses be free to serve who they want to?

> The logical conclusion of your arguments basically Fascism - putting profit above people

Fascism is about putting profit about people? I thought Fascism was commonly associated with National Socialism which doesn't sound very pro profits.

On the contrary the entire philosophy was built around the idea that every sector of society -- business, labor, churches, and so on should serve the end goals of state and more abstractly the nation/race. But today ignorant people use the term "fascism" to mean "something I don't like". It's a shanda.

"Extremists" are those who compel, nay blackmail, their corps. to stop selling to agencies and people whose politics and world views differ from theirs. 100% of law enforcement isn't in the business of caging children. Please stop making straw-mans - your security is also the responsibility of the agencies who you demonize as "child cagers". These same individuals will protect and serve you, while you take away their ability to use technology to do their jobs

ICE was specifically called out and have indeed been caging children - not a straw man. The people doing that neither serve nor protect me and should not have been given that “job to do” in the first place. Making something “a job” doesn’t suddenly make it ok. That’s just an argument from authority - you might want to look that term up.

Law enforcement in the broader sense is a different matter - there’s obviously legitimate use cases. But extreme caution needs to be exercised. Those folks do indeed have a valuable job to do - but within reason and never beyond reproach.

Also I don’t think you know what blackmail is. Now if I had a penny for every conservative boycott of some “liberal” company... I guess those folks are all “extremists”.

Please please explain how rekognition aids ICE in caging children. This is indeed a strawman, unless you prove a direct connection.

And then there's the deathly silence on actual evidence that a majority of Amazonians even want this. Instead, it's very likely the voice of a radical, but vocal minority, quite literally blackmailing the company.

Here's the blackmail clause - "deny ICE rekognition access, or we'll leak confidential company information to BuzzFeed (and the media) and make a hit piece of your policy in an echo chamber that sympathizes with our politics"

I could give a crap about conservatives protesting companies for politician reasons. I don't believe a baker should be able to deny a gay couple a wedding cake, and I certainly don't believe Amazon or valley intersectionaliats should have the power to deny the government of the United States, whose policies have checks and balances in 3 branches of governece, access to tech.

If, on the contrary, you proposed a legal challenge in the courts on constitutional merits, I'd support that. But this power trip of valley intersectionaliats de-platforming everything they disagree with is exhausting, in a free and mature democracy like the US.

> We really need center of the aisle, common sense to be popular again in this country

Okay, so left of the democrats.

I'd be fine with that.

I don't think you know what intersectionality is.

Not giving law enforcement broad surveillance capabilities hardly seems like a position conservatives would oppose. As usual the ones who yell loudest about tyranny are also the ones who cheerfully fight for it.

What tyranny and who's cheering for it? No one here wants anyone's constitutional rights violated. Pro tip - if there was brazen constitutional violations arising from this tech, Amazon lawyers would probably be the first to sue the government over it

This is all about the letf's power tripping political opponents out of a platform. It's a 180 phase shift of a scenario where some employers deny insurance coverage of birth control on religious or moral grounds.

I do not agree with either side on this behavior, and frankly find this a hole in the Constitution where a company can simply deny a legal transaction because they disagree with the recipient of the service

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact