Uge 4 i digitale medier, platforme og teknologi

Her har du den mest kom­plette opda­ter­ing på sociale medi­er og dig­i­tal udvikling i mediebranchen. Du bliv­er holdt opdateret og spar­er mass­er af tid.

Dig­i­tal Ugerevy udgives af mig, Lars K Jensen, og jeg arbe­jder til daglig med audi­ence-udvikling i Berlingske-kon­cer­nen.

Gør som de andre. Få overb­likket allerede i ind­bakken:

Du kan altid afmelde dig igen ved at bruge lin­ket i bun­den af hvert nyheds­brev

I denne uge: skjul

Del 1: Sociale medier 

Her find­er du his­to­ri­er om eller fra de store sociale plat­forme, der har en enorm rolle at spille i kon­tak­ten mellem udgiver/afsender og mod­tager.

👍 Meta vil lade EU-borgere frakoble Facebook- og Instagram-konti

Fra The Verge:

Under the changes, users will also be able to unbun­dle their Mar­ket­place and Face­book Gam­ing accounts, though in some cas­es Meta says this will lim­it fea­tures.

💡 Status på kampen mellem sociale og publicistiske medier

Hvis du man­gler et write-up på, hvad der er sket de sen­este måned­er og år i forhold til sociale medi­er og dig­i­tale medi­er, så har du det her.

Fra Kfo­rum:

Nyhedsme­dierne fyr­er, og annonce­sal­get er i frit fald. En af fork­laringerne er, at trafikken fra sociale medi­er er faldet med 50 pct. på blot tre år. 

🔀 Metas Oversight Board omgør beslutning med Holocaust-post

Fra hes­tens mund:

The Over­sight Board has over­turned Meta’s orig­i­nal deci­sion to leave up an Insta­gram post con­tain­ing false and dis­tort­ed claims about the Holo­caust. The Board finds that the con­tent vio­lat­ed Meta’s Hate Speech Com­mu­ni­ty Stan­dard, which bans Holo­caust denial.

😴 Instagram vil sende teenagere tidligere i seng

Fra Social Media Today:

Build­ing on its lat­est round of teen safe­ty fea­tures, Insta­gram has today announced new night­time nudges, which will prompt users to turn their damn phone off when they’re using it late at night.

Mere om Meta og de unge:

🧾 Nu kan du slippe for "læst"-kvitteringer i Instagram-DMs

Fra Social Media Today:

After announc­ing that it was com­ing last Novem­ber, Insta­gram has this week rolled out the capac­i­ty for users to switch off read receipts for their DMs, so you can con­trol when “Read” and “Seen” indi­ca­tors appear (or don’t) for the sender with­in mes­sage streams.

📼 TikTok tester 30 minutter i uploads

Fra Social Media Today:

As you can see in this exam­ple, shared by social media expert Matt Navar­ra, TikTok’s cur­rent­ly test­ing the new 30 minute upload option in the beta ver­sion of the app.

😬 Bliver 2024 et hårdt år for TikTok?

Det han­dler blandt andet om algo­rit­men.

Fra Digi­day:

TikTok’s algo­rithm has been the key rea­son for its suc­cess; draw­ing users in, and select­ing more con­tent for them to ogle over as they scroll their For You page. But as the say­ing goes, noth­ing lasts for­ev­er, and unfor­tu­nate­ly that’s the same for the platform’s algo­rithm.

Mere:

🤚 LinkedIn fjerner "lookalike" som målretning

Fra Social Media Today:

LinkedIn has announced that it’s remov­ing Looka­like Audi­ences as an ad tar­get­ing option, which could change the way that you look to reach your tar­get audi­ence in the app.

📞 X udruller lyd- og videokald til Android

Fra TechCrunch:

Elon Musk’s own social net­work X is rolling out the abil­i­ty to make audio and video calls direct­ly from the app to its Android client.

📄 Threads arbejder på kladdetilstand

Fra Social Media Today:

Accord­ing to a new back-end code dis­cov­ery in the Threads app, Threads users will soon be able to save their posts as drafts, enabling you to return and revise your mis­sives at anoth­er time.

🕵️‍♀️ Apps høster data via iPhone-notifikationer

Fra Giz­mo­do:

iPhone apps includ­ing Face­book, LinkedIn, Tik­Tok, and X/Twitter are skirt­ing Apple’s pri­va­cy rules to col­lect user data through noti­fi­ca­tions, accord­ing to tests by secu­ri­ty researchers at Mysk Inc., an app devel­op­ment com­pa­ny.

🤝 Udenom platformene: Flere creators går direkte til virksomhederne

Fra Digi­day:

Because of the plat­forms’ unpre­dictable algo­rithms and mys­te­ri­ous rev­enue shar­ing mod­els, social media is becom­ing more of a mar­ket­ing vehi­cle for cre­ators rather than a means to dri­ve rev­enue itself, cre­ators say. Instead of wait­ing on the cre­ator fund promised land, cre­ators are rethink­ing how they view the plat­forms, with some using social media as a means to get the atten­tion of brands rather than to gen­er­ate rev­enue.

Del 2: Digitale tendenser

Her kan du læse nogle af ugens ten­denser inden­for medi­er. Det kan være en ny metode eller til­gang, branchen forsøger at tage til sig.

✨ Så kom DR Medieforsknings årlige rapport

Du kan læse artik­lerne fra rap­porten hos DR — og hente den som PDF.

Og her er en række artik­ler om den:

🎧 Apple introducerer transkribering af alle podcasts

Fra Apple selv:

Apple auto­mat­i­cal­ly gen­er­ates tran­scripts after a new episode is pub­lished. Your episode will be avail­able for lis­ten­ing right away, and the tran­script will be avail­able short­ly after­wards. There will be a short delay while we process your tran­script. If por­tions of your episode change with dynam­i­cal­ly insert­ed audio, Apple Pod­casts will not dis­play the seg­ments of the audio that have changed since the orig­i­nal tran­scrip­tion. Music lyrics are also not dis­played in the tran­scripts.

🎓 Aftenposten deler erfaringer med lydartikler

Fra INMA:

In a recent blog post, we shared that we cre­at­ed a cloned voice using machine learn­ing, mak­ing it pos­si­ble for our audi­ence to lis­ten to arti­cles from Aften­posten. Using quan­ti­ta­tive data, we learned what peo­ple did — and didn’t — like.

🤦‍♂️ Google News prioriterede AI-skabte "nyheder"

Fra New York Post:

Google News search­es turn up AI-gen­er­at­ed arti­cles that brazen­ly steal from legit­i­mate media out­lets – and The Post has already iden­ti­fied at least one such ripoff of its own pub­lished work.

👋 Wikipedia lukker i Rusland

Fra Bloomberg:

A cam­paign to replace the country’s Wikipedia with a more pli­ant alter­na­tive seems near com­ple­tion. 

Del 3: Business

Her kan du læse nogle af ugens his­to­ri­er og ten­denser inden­for for­ret­ning­sud­vikling i mediebranchen. Alt­så det, der han­dler om at tjene penge.

⚽ Her er Ekstra Bladets planer med fodboldopkøb

🤐 Medier er forsigtige med at føje chatbots til OpenAI-butik

Fra Digi­day:

Ope­nAI launched an app store called GPT Store on Jan. 10 where devel­op­ers can share their cus­tom Chat­G­PT chat­bots pow­ered by gen­er­a­tive arti­fi­cial intel­li­gence, but dig­i­tal pub­lish­ing com­pa­nies haven’t been all that eager to build chat­bots — or add the chat­bots they’ve already built — and make them avail­able on OpenAI’s app store just yet.

Relateret:

🤖 Google åbner for beta-adgang til AI-tool til annoncører i UK og USA

Fra Digi­day:

The search giant yes­ter­day said its “con­ver­sa­tion­al” gen­er­a­tive AI tools are now avail­able for beta access to adver­tis­ers in the U.S. and U.K. before glob­al access arrives for Eng­lish-lan­guage adver­tis­ers in com­ing weeks.  

💰 Apple barsler med gebyrer og begrænsninger for "side-loading"

Fra The Wall Street Jour­nal:

Meta, Spo­ti­fy and oth­er com­pa­nies are weigh­ing new options for cus­tomers as Apple makes changes to com­ply with new Euro­pean law

Mere:

📈 Netflix når 260 mio. betalende

Fra Medi­aWatch:

Sid­ste år tjente stream­ing­gi­gan­ten Net­flix 5,4 mil­liarder dol­lar. Det svar­er til 37 mil­liarder kro­ner.

Del 4: Udvikling og tech

Her får du et kig på nogle af ugens his­to­ri­er inden­for den dig­i­tale udvikling og teknolo­gi, der enten direk­te eller indi­rek­te kan have betyd­ning for medier/udgivere.

😲 AI-genereret Taylor Swift-porno gik viralt

Fra 404 Media:

404 Media traced back the viral and abu­sive images of Tay­lor Swift to a spe­cif­ic Telegram com­mu­ni­ty. That group some­times uses Microsoft's AI tools.

Mere:

  • The Tay­lor Swift deep­fakes are a warn­ing [Plat­former]

✅ Blockchain-tool skal bekæmpe deepfakes

Fra Journalism.co.uk:

One com­pa­ny that is look­ing to help news­rooms tack­le the onslaught of deep fake imagery is Tai­wan-based Num­bers Pro­to­col. It devel­oped a tool that embeds meta­da­ta into a pho­to, which helps estab­lish where it came from, when it was shot and, ide­al­ly, infor­ma­tion about the author. This is under­pinned by blockchain tech­nol­o­gy, which allows view­ers to see the photo’s com­plete his­to­ry, includ­ing any alter­ations.

👉 USA: Højreorienterede medier er mere åbne overfor AI-bots

Fra Wired:

New data shows that over 88 per­cent of top-ranked news out­lets in the US now block web crawlers used by arti­fi­cial intel­li­gence com­pa­nies to col­lect train­ing data for chat­bots and oth­er AI projects. One sec­tor of the news busi­ness is a glar­ing out­lier, though: Right-wing media lags far behind their lib­er­al coun­ter­parts when it comes to bot-block­ing.

🌐 Google barsler med generative AI-elementer i Chrome

Fra Social Media Today:

Google has announced some new gen­er­a­tive AI ele­ments for Chrome, which will pro­vide auto­mat­ed assis­tance to improve your expe­ri­ence, along with artis­tic inspi­ra­tion in its web brows­er app.

✅ Kan ChatGPT faktatjekke?

Fra Thomas Bæk­dal:

Instead, it's look­ing up what we, as the press and as fact-check­ing sites, have already writ­ten about a top­ic, and then com­par­ing with that. In oth­er words, it's using our jour­nal­is­tic work. In fact, with­out real human jour­nal­ists first cov­er­ing these sto­ries, Chat­G­PT would have noth­ing to look at, and as such wouldn't even be able to do this.

😡 Britiske kunstnere overvejer at sagsøge AI-firmaer

Fra The Guardian:

British artists have con­tact­ed US lawyers to dis­cuss join­ing a class action against Mid­jour­ney and oth­er AI firms, while oth­ers have told the Observ­er that they may bring their own legal action in the UK.

🤔 Er AI fantastisk eller forbandelse for lokalmedier?

Fra Poyn­ter:

Here are a few ques­tions that could shape the out­come — clos­ing with a mod­est pro­pos­al for how AI com­pa­nies could be the heroes instead of the vil­lains in the sto­ry of local news.

🎧 Sådan arbejder podcast-netværk med AI-værktøjer

Fra Digi­day:

Pod­cast net­works, like Acast, iHeart­Media and Spo­ti­fy, are also test­ing these tools to increase their out­reach to prospec­tive clients, expand the range of shows that fit a buyer’s brief and trans­late shows into dif­fer­ent lan­guages.

💀 Nyt værktøj kan forgifte AI-modeller

Fra Ven­ture­Beat:

It’s here: months after it was first announced, Night­shade, a new, free soft­ware tool allow­ing artists to “poi­son” AI mod­els seek­ing to train on their works, is now avail­able for artists to down­load and use on any art­works they see fit.